row_id
string
section
string
section_slug
string
resource_type
string
marker
string
title
string
url
string
url_kind
string
domain
string
annotation
string
description
string
key_contribution
string
novelty
string
impact
string
signal
string
signal_strength
string
source_readme
string
source_line
int64
source_url
string
date_added
string
collection
string
collection_slug
string
user_goal
string
lifecycle_stages
string
audience
string
loop_layer
string
scope_fit
string
evidence_class
string
evidence_tier
string
source_status
string
canonical_url
string
source_title
string
source_description
string
authors
string
publication_date
string
publication_year
string
publication_venue
string
publisher
string
doi
string
publication_note
string
primary_category
string
metadata_source
string
github_repo
string
github_stars
string
github_forks
string
github_license
string
github_created_at
string
github_updated_at
string
arxiv_id
string
audited_at
timestamp[ms]
ale-0201
Model-Level Recurrence
model-level-recurrence
Paper
📄
LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation
https://iclr.cc/virtual/2026/poster/10009450
external
iclr.cc
Trains variable-length latent trajectories with time and step-size conditioning plus shortcut consistency, allowing one model to trade compute for quality across inference budgets without retraining.
Trains variable-length latent trajectories with time and step-size conditioning plus shortcut consistency, allowing one model to trade compute for quality across inference budgets without retraining.
Trains variable-length latent trajectories with time and step-size conditioning plus shortcut consistency, allowing one model to trade compute for quality across inference budgets without retraining.
Reuses learned computation inside one model inference rather than repeating a full agent run. Trains variable-length latent trajectories with time and step-size conditioning plus shortcut consistency, allowing one model to trade compute for quality across inference budgets without retraining.
Use LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
804
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L804
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;budget
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-paper
A
ok
https://iclr.cc/virtual/2026/poster/10009450
ICLR Poster LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation ICLR 2026
Ahmadreza Jeddi; Marco Ciccone; Babak Taati
2026
2026
International Conference on Learning Representations (ICLR)
International Conference on Learning Representations
Published at ICLR 2026; metadata verified from the official conference poster page.
ICLR proceedings
2026-08-13T14:24:04
ale-0202
Model-Level Recurrence
model-level-recurrence
Paper
📄
MoDr: Mixture-of-Depth-Recurrent Transformers for Test-Time Reasoning
https://iclr.cc/virtual/2026/poster/10011117
external
iclr.cc
Replaces a single recurrent reasoning path with dynamically routed LoRA branches, adding solution-space exploration and load-balanced routing to the Huginn-style depth-recurrent backbone.
Replaces a single recurrent reasoning path with dynamically routed LoRA branches, adding solution-space exploration and load-balanced routing to the Huginn-style depth-recurrent backbone.
Replaces a single recurrent reasoning path with dynamically routed LoRA branches, adding solution-space exploration and load-balanced routing to the Huginn-style depth-recurrent backbone.
Reuses learned computation inside one model inference rather than repeating a full agent run. Replaces a single recurrent reasoning path with dynamically routed LoRA branches, adding solution-space exploration and load-balanced routing to the Huginn-style depth-recurrent backbone.
Use MoDr: Mixture-of-Depth-Recurrent Transformers for Test-Time Reasoning to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
805
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L805
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-paper
A
ok
https://iclr.cc/virtual/2026/poster/10011117
ICLR Poster MoDr: Mixture-of-Depth-Recurrent Transformers for Test-Time Reasoning ICLR 2026
Xiaojing Zhang; Haifeng Wu; Gang He; Jiyang Shen; Bochen Lyu; Zhanxing Zhu
2026
2026
International Conference on Learning Representations (ICLR)
International Conference on Learning Representations
Published at ICLR 2026; metadata verified from the official conference poster page.
ICLR proceedings
2026-08-13T14:24:04
ale-0203
Model-Level Recurrence
model-level-recurrence
Paper
📄
ChainGPT: Dual-Reasoning Model with Recurrent Depth and Multi-Rank State Updates
https://iclr.cc/virtual/2026/poster/10007767
external
iclr.cc
Combines within-layer multi-substep state updates, state-guided sparse attention, across-layer recurrence, and adaptive stopping to increase latent reasoning depth without extending visible chain-of-thought.
Combines within-layer multi-substep state updates, state-guided sparse attention, across-layer recurrence, and adaptive stopping to increase latent reasoning depth without extending visible chain-of-thought.
Combines within-layer multi-substep state updates, state-guided sparse attention, across-layer recurrence, and adaptive stopping to increase latent reasoning depth without extending visible chain-of-thought.
Reuses learned computation inside one model inference rather than repeating a full agent run. Combines within-layer multi-substep state updates, state-guided sparse attention, across-layer recurrence, and adaptive stopping to increase latent reasoning depth without extending visible chain-of-thought.
Use ChainGPT: Dual-Reasoning Model with Recurrent Depth and Multi-Rank State Updates to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
806
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L806
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;state;exit
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-paper
A
ok
https://iclr.cc/virtual/2026/poster/10007767
ICLR Poster ChainGPT: Dual-Reasoning Model with Recurrent Depth and Multi-Rank State Updates ICLR 2026
Yunao Zheng; Xiaojie Wang; Lei Ren; Chen Wei
2026
2026
International Conference on Learning Representations (ICLR)
International Conference on Learning Representations
Published at ICLR 2026; metadata verified from the official conference poster page.
ICLR proceedings
2026-08-13T14:24:04
ale-0204
Model-Level Recurrence
model-level-recurrence
Paper
📄
Think-at-Hard: Selective Latent Iterations to Improve Reasoning Language Models
https://openreview.net/forum?id=eQaJSRZiGn
external
openreview.net
Learns when a token needs extra latent refinement, using a neural decider, depth-aware LoRA, and cross-iteration attention to avoid always paying for or being degraded by additional loops.
Learns when a token needs extra latent refinement, using a neural decider, depth-aware LoRA, and cross-iteration attention to avoid always paying for or being degraded by additional loops.
Learns when a token needs extra latent refinement, using a neural decider, depth-aware LoRA, and cross-iteration attention to avoid always paying for or being degraded by additional loops.
Reuses learned computation inside one model inference rather than repeating a full agent run. Learns when a token needs extra latent refinement, using a neural decider, depth-aware LoRA, and cross-iteration attention to avoid always paying for or being degraded by additional loops.
Use Think-at-Hard: Selective Latent Iterations to Improve Reasoning Language Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
807
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L807
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;budget
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-paper
A
ok
https://openreview.net/challenge?redirect=%2Fforum%3Fid%3DeQaJSRZiGn
Verifying your browser | OpenReview
Tianyu Fu; Yichen You; Zekai Chen; Guohao Dai; Huazhong Yang; Yu Wang
2026
2026
International Conference on Machine Learning (ICML)
OpenReview
Published at ICML 2026; venue and authors verified from the official OpenReview record.
OpenReview
2026-08-13T14:24:04
ale-0205
Model-Level Recurrence
model-level-recurrence
Paper
📄
Fixed-Point Reasoners: Stable and Adaptive Deep Looped Transformers
https://arxiv.org/abs/2606.18206
external
arxiv.org
Stabilizes very deep recurrence with pre-normalization and residual scaling, then uses latent-state convergence as the halting signal so the model allocates more iterations to harder Sudoku, maze, state-tracking, and ARC-AGI instances without a separate stopping head.
Stabilizes very deep recurrence with pre-normalization and residual scaling, then uses latent-state convergence as the halting signal so the model allocates more iterations to harder Sudoku, maze, state-tracking, and ARC-AGI instances without a separate stopping head.
Stabilizes very deep recurrence with pre-normalization and residual scaling, then uses latent-state convergence as the halting signal so the model allocates more iterations to harder Sudoku, maze, state-tracking, and ARC-AGI instances without a separate stopping head.
Reuses learned computation inside one model inference rather than repeating a full agent run. Stabilizes very deep recurrence with pre-normalization and residual scaling, then uses latent-state convergence as the halting signal so the model allocates more iterations to harder Sudoku, maze, state-tracking, and ARC-AGI i...
Use Fixed-Point Reasoners: Stable and Adaptive Deep Looped Transformers to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2606.18206; inspect its method and evaluation before treating results as production evidence.
medium
README.md
808
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L808
2026-07-20
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;state;exit
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-paper
A
ok
https://openreview.net/pdf/51350b6e425ed0500ac9eb9cec78ba15d9f5d1ba.pdf
[2606.18206] Fixed-Point Reasoners: Stable and Adaptive Deep Looped Transformers
Looped architectures provide an inductive bias toward learning step-by-step procedures for tasks that require compositional reasoning. The number of effective layers reached by looping determines the quality of the solution these models find. Like deep architectures, looped architectures are prone to a signal propagati...
Sajad Movahedi; Vera Milovanović; Shlomo Libo Feigin; Alexander Theus; Thomas Hofmann; Valentina Boeva; T. Konstantin Rusch; Antonio Orvieto
2026
2026
Proceedings of the 43rd International Conference on Machine Learning (ICML), PMLR 306
PMLR
Published in Proceedings of the 43rd International Conference on Machine Learning (ICML), PMLR 306; the linked arXiv record remains available for open access.
cs.AI
PMLR camera-ready record
2606.18206
2026-08-13T14:24:04
ale-0206
Model-Level Recurrence
model-level-recurrence
Paper
📄
Loop the Loopies!
https://arxiv.org/abs/2607.16051
external
arxiv.org
Introduces the Loopie family of sparse looped Transformers (20B parameters with 2B active and 6B with 0.6B active); the authors' matched-compute experiments over 3.5T pretraining tokens overtake a reproduced vanilla 30B-A3B baseline after roughly 600B tokens, while post-training plus test-time scaling reaches 35/42 on ...
Introduces the Loopie family of sparse looped Transformers (20B parameters with 2B active and 6B with 0.6B active); the authors' matched-compute experiments over 3.5T pretraining tokens overtake a reproduced vanilla 30B-A3B baseline after roughly 600B tokens, while post-training plus test-time scaling reaches 35/42 on ...
Introduces the Loopie family of sparse looped Transformers (20B parameters with 2B active and 6B with 0.6B active); the authors' matched-compute experiments over 3.5T pretraining tokens overtake a reproduced vanilla 30B-A3B baseline after roughly 600B tokens, while post-training plus test-time scaling reaches 35/42 on ...
Reuses learned computation inside one model inference rather than repeating a full agent run. Introduces the Loopie family of sparse looped Transformers (20B parameters with 2B active and 6B with 0.6B active); the authors' matched-compute experiments over 3.5T pretraining tokens overtake a reproduced vanilla 30B-A3B ba...
Use Loop the Loopies! to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2607.16051; inspect its method and evaluation before treating results as production evidence.
medium
README.md
814
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L814
2026-07-20
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification;budget
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2607.16051
[2607.16051] Loop the Loopies!
We present the Loopie series, consisting of two Mixture-of-Experts (MoE) models: a 20B-parameter model with 2B active parameters and a 6B-parameter model with 0.6B active parameters. Looped Transformers have long faced a challenge: given an N times increase in pre-training compute, increasing the parameter count by a f...
Zitian Gao; Yilong Chen; Yihao Xiao; Xinyu Yang; Ran Tao; Joey Zhou; Bryan Dai
2026-07-17
2026
arXiv
arXiv
cs.CL
arxiv-api
2607.16051
2026-08-13T14:24:04
ale-0207
Model-Level Recurrence
model-level-recurrence
Paper
📄
LoopMoE: Unifying Iterative Computation with Mixture-of-Experts for Language Modeling
https://arxiv.org/abs/2606.04438
external
arxiv.org
Combines sparse expert routing with weight-shared recurrence through iteration-conditioned adaptive normalization and capacity balancing; under matched parameters, per-token FLOPs, and active-sublayer ratios, the reported 3B model beats its vanilla MoE control on 8 of 9 benchmarks by more than one point on average, wit...
Combines sparse expert routing with weight-shared recurrence through iteration-conditioned adaptive normalization and capacity balancing; under matched parameters, per-token FLOPs, and active-sublayer ratios, the reported 3B model beats its vanilla MoE control on 8 of 9 benchmarks by more than one point on average, wit...
Combines sparse expert routing with weight-shared recurrence through iteration-conditioned adaptive normalization and capacity balancing; under matched parameters, per-token FLOPs, and active-sublayer ratios, the reported 3B model beats its vanilla MoE control on 8 of 9 benchmarks by more than one point on average, wit...
Reuses learned computation inside one model inference rather than repeating a full agent run. Combines sparse expert routing with weight-shared recurrence through iteration-conditioned adaptive normalization and capacity balancing; under matched parameters, per-token FLOPs, and active-sublayer ratios, the reported 3B m...
Use LoopMoE: Unifying Iterative Computation with Mixture-of-Experts for Language Modeling to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2606.04438; inspect its method and evaluation before treating results as production evidence.
medium
README.md
815
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L815
2026-07-20
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification;budget
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2606.04438
[2606.04438] LoopMoE: Unifying Iterative Computation with Mixture-of-Experts for Language Modeling
Mixture-of-Experts (MoE) and looped architectures scale models along two orthogonal axes, namely parameter capacity and effective depth. However, mainstream looped architectures rely on dense backbones that couple parameter count with per-token FLOPs, which makes it impossible to isolate the effect of iterative computa...
Wenkai Chen; Tianshu Li; Wenyong Huang; Yichun Yin; Lifeng Shang; Chengwei Qin
2026-06-03
2026
arXiv
arXiv
cs.LG
arxiv-api
2606.04438
2026-08-13T14:24:04
ale-0208
Model-Level Recurrence
model-level-recurrence
Paper
📄
Sparse Layers are Critical to Scaling Looped Language Models
https://arxiv.org/abs/2605.09165
external
arxiv.org
Finds that dense looped models lose their scaling advantage while looped MoE models recover expressivity because routing diverges across repeated passes; also shows that loop boundaries form stronger early-exit points than arbitrary layers in standard Transformers.
Finds that dense looped models lose their scaling advantage while looped MoE models recover expressivity because routing diverges across repeated passes; also shows that loop boundaries form stronger early-exit points than arbitrary layers in standard Transformers.
Finds that dense looped models lose their scaling advantage while looped MoE models recover expressivity because routing diverges across repeated passes; also shows that loop boundaries form stronger early-exit points than arbitrary layers in standard Transformers.
Reuses learned computation inside one model inference rather than repeating a full agent run. Finds that dense looped models lose their scaling advantage while looped MoE models recover expressivity because routing diverges across repeated passes; also shows that loop boundaries form stronger early-exit points than arb...
Use Sparse Layers are Critical to Scaling Looped Language Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2605.09165; inspect its method and evaluation before treating results as production evidence.
medium
README.md
816
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L816
2026-07-20
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;exit
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2605.09165
[2605.09165] Sparse Layers are Critical to Scaling Looped Language Models
Looped language models repeat a set of transformer layers through depth, reducing memory costs and providing natural early-exit points at loop boundaries. However, looped models do not scale as favorably as standard transformers with unique layers. We compare standard and Mixture-of-Experts (MoE) transformers, with and...
Ryan Lee; Jacob Biloki; Edward J. Hu; Jonathan May
2026-05-09
2026
arXiv
arXiv
cs.LG
arxiv-api
2605.09165
2026-08-13T14:24:04
ale-0209
Model-Level Recurrence
model-level-recurrence
Paper
📄
A Mechanistic Analysis of Looped Reasoning Language Models
https://arxiv.org/abs/2604.11791
external
arxiv.org
Traces cyclic recurrence inside looped language models and finds that many studied layers converge to distinct fixed points, with attention behavior stabilizing as recurrent blocks repeat feedforward-like inference stages; tests how block size, input injection, and normalization shape those dynamics.
Traces cyclic recurrence inside looped language models and finds that many studied layers converge to distinct fixed points, with attention behavior stabilizing as recurrent blocks repeat feedforward-like inference stages; tests how block size, input injection, and normalization shape those dynamics.
Traces cyclic recurrence inside looped language models and finds that many studied layers converge to distinct fixed points, with attention behavior stabilizing as recurrent blocks repeat feedforward-like inference stages; tests how block size, input injection, and normalization shape those dynamics.
Reuses learned computation inside one model inference rather than repeating a full agent run. Traces cyclic recurrence inside looped language models and finds that many studied layers converge to distinct fixed points, with attention behavior stabilizing as recurrent blocks repeat feedforward-like inference stages; tes...
Use A Mechanistic Analysis of Looped Reasoning Language Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2604.11791; inspect its method and evaluation before treating results as production evidence.
medium
README.md
817
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L817
2026-07-20
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2604.11791
[2604.11791] A Mechanistic Analysis of Looped Reasoning Language Models
Reasoning has become a central capability in large language models. Recent research has shown that reasoning performance can be improved by looping an LLM's layers in the latent dimension, resulting in looped reasoning language models. Despite promising results, few works have investigated how their internal dynamics d...
Hugh Blayney; Álvaro Arroyo; Johan Obando-Ceron; Pablo Samuel Castro; Aaron Courville; Michael M. Bronstein; Xiaowen Dong
2026-04-13
2026
arXiv
arXiv
39 pages, 63 figures
cs.LG
arxiv-api
2604.11791
2026-08-13T14:24:04
ale-0210
Model-Level Recurrence
model-level-recurrence
Paper
📄
Parcae: Scaling Laws For Stable Looped Language Models
https://openreview.net/forum?id=ri0LAMdhd9
external
openreview.net
Treats the recurrent block as a dynamical system, constrains its stability, and studies how training and test-time recurrence trade parameters for additional computation.
Treats the recurrent block as a dynamical system, constrains its stability, and studies how training and test-time recurrence trade parameters for additional computation.
Treats the recurrent block as a dynamical system, constrains its stability, and studies how training and test-time recurrence trade parameters for additional computation.
Reuses learned computation inside one model inference rather than repeating a full agent run. Treats the recurrent block as a dynamical system, constrains its stability, and studies how training and test-time recurrence trade parameters for additional computation.
Use Parcae: Scaling Laws For Stable Looped Language Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
818
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L818
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-paper
A
ok
https://openreview.net/challenge?redirect=%2Fforum%3Fid%3Dri0LAMdhd9
Verifying your browser | OpenReview
Hayden Prairie; Zachary Novack; Taylor Berg-Kirkpatrick; Daniel Y. Fu
2026
2026
Learning to Iterate Workshop at ICLR 2026
OpenReview
Workshop paper at the Learning to Iterate Workshop at ICLR 2026; not an ICLR main-conference paper.
OpenReview
2026-08-13T14:24:04
ale-0211
Model-Level Recurrence
model-level-recurrence
Paper
📄
SpiralFormer: Looped Transformers Can Learn Hierarchical Dependencies via Multi-Resolution Recursion
https://arxiv.org/abs/2602.11698
external
arxiv.org
Recurs over progressively compressed representations so repeated layers specialize across resolutions instead of recomputing every token at full resolution.
Recurs over progressively compressed representations so repeated layers specialize across resolutions instead of recomputing every token at full resolution.
Recurs over progressively compressed representations so repeated layers specialize across resolutions instead of recomputing every token at full resolution.
Reuses learned computation inside one model inference rather than repeating a full agent run. Recurs over progressively compressed representations so repeated layers specialize across resolutions instead of recomputing every token at full resolution.
Use SpiralFormer: Looped Transformers Can Learn Hierarchical Dependencies via Multi-Resolution Recursion to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2602.11698; inspect its method and evaluation before treating results as production evidence.
medium
README.md
819
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L819
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;budget
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2602.11698
[2602.11698] SpiralFormer: Looped Transformers Can Learn Hierarchical Dependencies via Multi-Resolution Recursion
Recursive (looped) Transformers decouple computational depth from parameter depth by repeatedly applying shared layers, providing an explicit architectural primitive for iterative refinement and latent reasoning. However, early looped Transformers often underperform non-recursive baselines of equal compute. While recen...
Chengting Yu; Xiaobo Shu; Yadao Wang; Yizhen Zhang; Haoyi Wu; You Wu; Rujiao Long; Ziheng Chen; Yuchi Xu; Wenbo Su; Bo Zheng
2026-02-12
2026
arXiv
arXiv
cs.LG
arxiv-api
2602.11698
2026-08-13T14:24:04
ale-0212
Model-Level Recurrence
model-level-recurrence
Paper
📄
Training-Free Looped Transformers
https://arxiv.org/abs/2605.23872
external
arxiv.org
Retrofits recurrence onto frozen pretrained models by applying a damped mid-stack block as smaller refinement steps, testing when inference-time looping helps without fine-tuning.
Retrofits recurrence onto frozen pretrained models by applying a damped mid-stack block as smaller refinement steps, testing when inference-time looping helps without fine-tuning.
Retrofits recurrence onto frozen pretrained models by applying a damped mid-stack block as smaller refinement steps, testing when inference-time looping helps without fine-tuning.
Reuses learned computation inside one model inference rather than repeating a full agent run. Retrofits recurrence onto frozen pretrained models by applying a damped mid-stack block as smaller refinement steps, testing when inference-time looping helps without fine-tuning.
Use Training-Free Looped Transformers to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2605.23872; inspect its method and evaluation before treating results as production evidence.
medium
README.md
820
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L820
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2605.23872
[2605.23872] Training-Free Looped Transformers
We introduce training-free looped transformers, in which a lightweight inference-time wrapper loops a contiguous mid-stack block of layers of a frozen checkpoint without additional fine-tuning, continued training, or architectural changes. Unlike prior looped transformer methods that train with the looped structure end...
Lizhang Chen; Jonathan Li; Chen Liang; Ni Lao; Qiang Liu
2026-05-22
2026
arXiv
arXiv
cs.LG
arxiv-api
2605.23872
2026-08-13T14:24:04
ale-0213
Model-Level Recurrence
model-level-recurrence
Paper
📄
Loop, Think, & Generalize: Implicit Reasoning in Recurrent-Depth Transformers
https://arxiv.org/abs/2604.07822
external
arxiv.org
Shows systematic generalization and depth extrapolation in controlled recurrent-depth reasoning tasks while documenting overthinking when recurrence exceeds the useful computation horizon.
Shows systematic generalization and depth extrapolation in controlled recurrent-depth reasoning tasks while documenting overthinking when recurrence exceeds the useful computation horizon.
Shows systematic generalization and depth extrapolation in controlled recurrent-depth reasoning tasks while documenting overthinking when recurrence exceeds the useful computation horizon.
Reuses learned computation inside one model inference rather than repeating a full agent run. Shows systematic generalization and depth extrapolation in controlled recurrent-depth reasoning tasks while documenting overthinking when recurrence exceeds the useful computation horizon.
Use Loop, Think, & Generalize: Implicit Reasoning in Recurrent-Depth Transformers to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2604.07822; inspect its method and evaluation before treating results as production evidence.
medium
README.md
821
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L821
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2604.07822
[2604.07822] Loop, Think, & Generalize: Implicit Reasoning in Recurrent-Depth Transformers
We study implicit reasoning, i.e. the ability to combine knowledge or rules within a single forward pass. While transformer-based large language models store substantial factual knowledge and rules, they often fail to compose this knowledge for implicit multi-hop reasoning, suggesting a lack of compositional generaliza...
Harsh Kohli; Srinivasan Parthasarathy; Huan Sun; Yuekun Yao
2026-04-09
2026
arXiv
arXiv
21 pages, 21 figures. Accepted at COLM 2026
cs.CL
arxiv-api
2604.07822
2026-08-13T14:24:04
ale-0214
Model-Level Recurrence
model-level-recurrence
Paper
📄
DeepLoop: Depth Scaling for Looped Transformers
https://arxiv.org/abs/2607.13491
external
arxiv.org
Derives residual-scaling rules that account for repeated parameter visits and tests them on GPT-style looped language models, addressing instability that nominal layer depth alone misses.
Derives residual-scaling rules that account for repeated parameter visits and tests them on GPT-style looped language models, addressing instability that nominal layer depth alone misses.
Derives residual-scaling rules that account for repeated parameter visits and tests them on GPT-style looped language models, addressing instability that nominal layer depth alone misses.
Reuses learned computation inside one model inference rather than repeating a full agent run. Derives residual-scaling rules that account for repeated parameter visits and tests them on GPT-style looped language models, addressing instability that nominal layer depth alone misses.
Use DeepLoop: Depth Scaling for Looped Transformers to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2607.13491; inspect its method and evaluation before treating results as production evidence.
medium
README.md
822
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L822
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2607.13491
[2607.13491] DeepLoop: Depth Scaling for Looped Transformers
Looped Transformers scale sequential computation by applying a compact stack of physical blocks for multiple rounds, increasing unrolled depth without increasing stored parameters. This reuse changes the residual-scaling problem: in an untied Transformer, each residual branch receives and applies its own parameter upda...
Shuzhen Li; Yifan Zhang; Jiacheng Guo; Quanquan Gu; Mengdi Wang
2026-07-15
2026
arXiv
arXiv
25 pages
cs.LG
arxiv-api
2607.13491
2026-08-13T14:24:04
ale-0215
Model-Level Recurrence
model-level-recurrence
Paper
📄
How Much Is One Recurrence Worth? Iso-Depth Scaling Laws for Looped Language Models
https://arxiv.org/abs/2604.21106
external
arxiv.org
Separates recurrent passes from unique parameter depth and training compute to estimate when an additional loop is worth more than adding distinct layers.
Separates recurrent passes from unique parameter depth and training compute to estimate when an additional loop is worth more than adding distinct layers.
Separates recurrent passes from unique parameter depth and training compute to estimate when an additional loop is worth more than adding distinct layers.
Reuses learned computation inside one model inference rather than repeating a full agent run. Separates recurrent passes from unique parameter depth and training compute to estimate when an additional loop is worth more than adding distinct layers.
Use How Much Is One Recurrence Worth? Iso-Depth Scaling Laws for Looped Language Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2604.21106; inspect its method and evaluation before treating results as production evidence.
medium
README.md
823
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L823
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2604.21106
[2604.21106] How Much Is One Recurrence Worth? Iso-Depth Scaling Laws for Looped Language Models
We measure how much one recurrence is worth to a looped (depth-recurrent) transformer, in equivalent unique parameters. From an iso-depth pretraining sweep across recurrence counts $r \in \{1, 2, 4, 8\}$ spanning ${\sim}50\times$ in training compute, we fit a joint scaling law $L = E + A\,(N_\text{once} + r^{\varphi} N...
Kristian Schwethelm; Daniel Rueckert; Georgios Kaissis
2026-04-22
2026
arXiv
arXiv
v3: substantially refined framing + minor corrections v2: added case studies on truncated-BPTT and hyperconnections
cs.LG
arxiv-api
2604.21106
2026-08-13T14:24:04
ale-0216
Model-Level Recurrence
model-level-recurrence
Paper
📄
LoopCoder: Scaling Code Intelligence via Looped Language Models
https://aclanthology.org/2026.findings-acl.796/
external
aclanthology.org
Scales looped code models through dense-to-loop initialization, recurrent pretraining, and post-training; the Findings of ACL 2026 paper reports a 40B-active/80B-total model trained on more than 12T code and general tokens.
Scales looped code models through dense-to-loop initialization, recurrent pretraining, and post-training; the Findings of ACL 2026 paper reports a 40B-active/80B-total model trained on more than 12T code and general tokens.
Scales looped code models through dense-to-loop initialization, recurrent pretraining, and post-training; the Findings of ACL 2026 paper reports a 40B-active/80B-total model trained on more than 12T code and general tokens.
Reuses learned computation inside one model inference rather than repeating a full agent run. Scales looped code models through dense-to-loop initialization, recurrent pretraining, and post-training; the Findings of ACL 2026 paper reports a 40B-active/80B-total model trained on more than 12T code and general tokens.
Use LoopCoder: Scaling Code Intelligence via Looped Language Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
829
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L829
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;budget
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-paper
A
ok
https://aclanthology.org/2026.findings-acl.796/
LoopCoder: Scaling Code Intelligence via Looped Language Models - ACL Anthology
Jian Yang, Wei Zhang, Shuyue Guo, Yizhi LI, Linzheng Chai, Zhengmao Ye, Shukai Liu, Yuyang Song, Jiajun Wu, Che Liu, Tianyu Zheng, Siwei Wu, Leo L, Xudong Ma, Chuan Hao, Ran Tao, Yan Xing, Jianzhou Wang, Mingjie Tang, Aishan Liu, Zhoujun Li, Xianglong Liu, Weifeng Lv, Bryan Dai. Findings of the Association for Computat...
Jian Yang; Wei Zhang; Shuyue Guo; Yizhi Li; Linzheng Chai; Zhengmao Ye; Shukai Liu; Yuyang Song; Jiajun Wu; Che Liu; Tianyu Zheng; Siwei Wu; Leo L; Xudong Ma; Chuan Hao; Ran Tao; Yan Xing; Jianzhou Wang; Mingjie Tang; Aishan Liu; Zhoujun Li; Xianglong Liu; Weifeng Lv; Bryan Dai
2026
2026
Findings of the Association for Computational Linguistics: ACL 2026
ACL Anthology
10.18653/v1/2026.findings-acl.796
html-meta
2026-08-13T14:24:04
ale-0217
Model-Level Recurrence
model-level-recurrence
Paper
📄
Looped World Models
https://arxiv.org/abs/2606.18208
external
arxiv.org
Introduces LoopWM, which repeatedly refines action-conditioned latent environment states with a parameter-shared Transformer core, stability constraints, and adaptive early exit for long-horizon simulation.
Introduces LoopWM, which repeatedly refines action-conditioned latent environment states with a parameter-shared Transformer core, stability constraints, and adaptive early exit for long-horizon simulation.
Introduces LoopWM, which repeatedly refines action-conditioned latent environment states with a parameter-shared Transformer core, stability constraints, and adaptive early exit for long-horizon simulation.
Reuses learned computation inside one model inference rather than repeating a full agent run. Introduces LoopWM, which repeatedly refines action-conditioned latent environment states with a parameter-shared Transformer core, stability constraints, and adaptive early exit for long-horizon simulation.
Use Looped World Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2606.18208; inspect its method and evaluation before treating results as production evidence.
medium
README.md
830
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L830
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;exit
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2606.18208
[2606.18208] Looped World Models
Current world models face a fundamental tension: faithful long-horizon simulation demands deep computation, but deeper models are expensive to deploy and prone to compounding errors. We resolve this by introducing Looped World Models (LoopWM), which are the first looped architectures for world modelling. Our method ite...
Hongyuan Adam Lu; Z. L. Victor Wei; Qun Zhang; Jinrui Zeng; Bowen Cao; Lingwei Meng; Mocheng Li; Zezhong Wang; Haonan Yin; Naifu Xue; Minyu Chen; Cenyuan Zhang; Zefan Zhang; Hao Wei; Jiawei Zhou; Haoran Xu; Hao Yang; Ronglai Zuo; Tongda Xu; Yonghao Li; Jian Chen; Hebin Wang; Zeyu Gao; Yang Li; Wei Zhao; Qimin Zhong; Si...
2026-06-16
2026
arXiv
arXiv
Technical Report
cs.LG
arxiv-api
2606.18208
2026-08-13T14:24:04
ale-0218
Model-Level Recurrence
model-level-recurrence
Paper
📄
Bridging the Gap Between Latent and Explicit Reasoning with Looped Transformers
https://arxiv.org/abs/2606.31779
external
arxiv.org
Introduces LOTUS, which loops over parallel latent thought blocks with intermediate supervision and reports explicit-chain-of-thought-level quality at 3B scale with lower thought-phase latency.
Introduces LOTUS, which loops over parallel latent thought blocks with intermediate supervision and reports explicit-chain-of-thought-level quality at 3B scale with lower thought-phase latency.
Introduces LOTUS, which loops over parallel latent thought blocks with intermediate supervision and reports explicit-chain-of-thought-level quality at 3B scale with lower thought-phase latency.
Reuses learned computation inside one model inference rather than repeating a full agent run. Introduces LOTUS, which loops over parallel latent thought blocks with intermediate supervision and reports explicit-chain-of-thought-level quality at 3B scale with lower thought-phase latency.
Use Bridging the Gap Between Latent and Explicit Reasoning with Looped Transformers to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2606.31779; inspect its method and evaluation before treating results as production evidence.
medium
README.md
831
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L831
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2606.31779
[2606.31779] Bridging the Gap Between Latent and Explicit Reasoning with Looped Transformers
Language models typically reason via explicit chain-of-thought (CoT), generating intermediate steps token-by-token. Latent CoT offers an alternative: it performs multi-step reasoning in the model's hidden states, replacing decoded tokens with continuous representations for greater efficiency. However, existing latent C...
Ying Fan; Anej Svete; Kangwook Lee
2026-06-30
2026
arXiv
arXiv
cs.LG
arxiv-api
2606.31779
2026-08-13T14:24:04
ale-0219
Model-Level Recurrence
model-level-recurrence
Paper
📄
Looped SSMs: Depth-Recurrence and Input Reshaping for Time Series Classification
https://arxiv.org/abs/2605.16048
external
arxiv.org
Extends tied-depth recurrence beyond Transformers by repeatedly applying a shared state-space block and reshaping inputs across iterations, testing which loop principles transfer across model families.
Extends tied-depth recurrence beyond Transformers by repeatedly applying a shared state-space block and reshaping inputs across iterations, testing which loop principles transfer across model families.
Extends tied-depth recurrence beyond Transformers by repeatedly applying a shared state-space block and reshaping inputs across iterations, testing which loop principles transfer across model families.
Reuses learned computation inside one model inference rather than repeating a full agent run. Extends tied-depth recurrence beyond Transformers by repeatedly applying a shared state-space block and reshaping inputs across iterations, testing which loop principles transfer across model families.
Use Looped SSMs: Depth-Recurrence and Input Reshaping for Time Series Classification to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2605.16048; inspect its method and evaluation before treating results as production evidence.
medium
README.md
832
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L832
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;verification;state
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2605.16048
[2605.16048] Looped SSMs: Depth-Recurrence and Input Reshaping for Time Series Classification
State Space Models (SSMs) are inherently recurrent along the sequence dimension, yet depth-recurrence - reusing the same block repeatedly across layers, as recently applied in looped transformers - has not been explored in this model family. We show that a looped SSM with $k$ parameters iterated $L$ times consistently ...
Mónika Farsang; Ramin Hasani; Daniela Rus; Radu Grosu
2026-05-15
2026
arXiv
arXiv
cs.LG
arxiv-api
2605.16048
2026-08-13T14:24:04
ale-0220
Model-Level Recurrence
model-level-recurrence
Paper
📄
Looped Diffusion Language Models
https://arxiv.org/abs/2605.26106
external
arxiv.org
Applies selective shared-layer recurrence to masked diffusion language models so effective depth can scale during training and inference without adding parameters.
Applies selective shared-layer recurrence to masked diffusion language models so effective depth can scale during training and inference without adding parameters.
Applies selective shared-layer recurrence to masked diffusion language models so effective depth can scale during training and inference without adding parameters.
Reuses learned computation inside one model inference rather than repeating a full agent run. Applies selective shared-layer recurrence to masked diffusion language models so effective depth can scale during training and inference without adding parameters.
Use Looped Diffusion Language Models to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2605.26106; inspect its method and evaluation before treating results as production evidence.
medium
README.md
833
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L833
2026-07-18
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2605.26106
[2605.26106] Looped Diffusion Language Models
Masked diffusion models (MDMs) have emerged as a promising alternative to autoregressive models for language modeling, yet the effective design of transformer architectures for MDMs remains underexplored. In this paper, we show that selectively looping the early-middle transformer layers significantly improves both tra...
Sanghyun Lee; Chunsan Hong; Seungryong Kim; Jonghyun Lee; Jongho Park; Dongmin Park
2026-05-25
2026
arXiv
arXiv
23 pages
cs.LG
arxiv-api
2605.26106
2026-08-13T14:24:04
ale-0221
Model-Level Recurrence
model-level-recurrence
Paper
📄
Metis: Memory Foundation Model
https://arxiv.org/abs/2607.26760
external
arxiv.org
Argues for native memory inside the backbone rather than bolted-on external modules: a persistent memory state compresses history, is read through memory attention, and is maintained gradient-free with forward passes only at inference. The model-level answer to problems the harness-level memory stacks keep patching.
Argues for native memory inside the backbone rather than bolted-on external modules: a persistent memory state compresses history, is read through memory attention, and is maintained gradient-free with forward passes only at inference. The model-level answer to problems the harness-level memory stacks keep patching.
Argues for native memory inside the backbone rather than bolted-on external modules: a persistent memory state compresses history, is read through memory attention, and is maintained gradient-free with forward passes only at inference. The model-level answer to problems the harness-level memory stacks keep patching.
Reuses learned computation inside one model inference rather than repeating a full agent run. Argues for native memory inside the backbone rather than bolted-on external modules: a persistent memory state compresses history, is read through memory attention, and is maintained gradient-free with forward passes only at i...
Use Metis: Memory Foundation Model to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2607.26760; inspect its method and evaluation before treating results as production evidence.
medium
README.md
834
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L834
2026-07-30
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;context;state
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2607.26760
[2607.26760] Metis: Memory Foundation Model
Recent advances in AI agents have increasingly internalized native capabilities into their underlying foundation models, giving rise to multimodal foundation models and large reasoning models. However, agent memory is still primarily implemented through external modules, leaving the native memory capability largely une...
Zeyu Zhang; Ziliang Guo; Yihang Sun; Xichong Zhang; Xixuan Hao; Zehao Lin; Yang Zhang; Xiaoyan Zhao; Tong Shen; Bo Tang; Zhi-Qin John Xu; Junchi Yan; Haofen Wang; Xu Chen; Feiyu Xiong; Zhiyu Li; Tat-Seng Chua
2026-07-29
2026
arXiv
arXiv
46 pages, 11 figures, 16 tables
cs.CL
arxiv-api
2607.26760
2026-08-13T14:24:04
ale-0222
Model-Level Recurrence
model-level-recurrence
Paper
📄
Memory Decoder at Scale: A Pretrained, Parametric Long-Term Memory
https://arxiv.org/abs/2607.27919
external
arxiv.org
Attacks the memory problem one layer below retrieval: decoder-only models entangle long-term memory and reasoning in a single parameter set, so memory capacity cannot be scaled independently of the reasoning capacity you are paying for.
Attacks the memory problem one layer below retrieval: decoder-only models entangle long-term memory and reasoning in a single parameter set, so memory capacity cannot be scaled independently of the reasoning capacity you are paying for.
Attacks the memory problem one layer below retrieval: decoder-only models entangle long-term memory and reasoning in a single parameter set, so memory capacity cannot be scaled independently of the reasoning capacity you are paying for.
Reuses learned computation inside one model inference rather than repeating a full agent run. Attacks the memory problem one layer below retrieval: decoder-only models entangle long-term memory and reasoning in a single parameter set, so memory capacity cannot be scaled independently of the reasoning capacity you are p...
Use Memory Decoder at Scale: A Pretrained, Parametric Long-Term Memory to assess inner latent computation as a model capability inside a separately governed agent loop.
Research source arXiv:2607.27919; inspect its method and evaluation before treating results as production evidence.
medium
README.md
835
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L835
2026-08-07
Learn
learn
Understand how recurrent model computation can power, but not replace, a governed agent loop.
act;context
researcher;evaluator;model-builder;agent-builder
model
adjacent
research-preprint
A
ok
https://arxiv.org/abs/2607.27919
[2607.27919] Memory Decoder at Scale: A Pretrained, Parametric Long-Term Memory
Decoder-only language models entangle long-term memory and reasoning in a single parameter set, making it difficult to scale memory capacity independently. Memory Decoder introduces a parametric long-term memory module but only studies it at a relatively small scale. In this work, we present Memory Decoder at Scale, sc...
Rubin Wei; Jiaqi Cao; Jiarui Wang; Junming Zhang; Qipeng Guo; Bowen Zhou; Zhouhan Lin
2026-07-30
2026
arXiv
arXiv
cs.CL
arxiv-api
2607.27919
2026-08-13T14:24:04
ale-0223
Agent Workflow Patterns
agent-workflow-patterns
Tool
🧰
Discovery Loop
https://www.discoveryloop.com/
external
www.discoveryloop.com
Company thesis for automating the experimental loop itself: propose, implement, run, examine, iterate, driven by frontier models over large-scale compute so thousands of experiments run in parallel instead of sequentially by hand. Starts with machine learning research and uses its own stack as the first customer. A sta...
Company thesis for automating the experimental loop itself: propose, implement, run, examine, iterate, driven by frontier models over large-scale compute so thousands of experiments run in parallel instead of sequentially by hand. Starts with machine learning research and uses its own stack as the first customer. A sta...
Company thesis for automating the experimental loop itself: propose, implement, run, examine, iterate, driven by frontier models over large-scale compute so thousands of experiments run in parallel instead of sequentially by hand. Starts with machine learning research and uses its own stack as the first customer. A sta...
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Company thesis for automating the experimental loop itself: propose, implement, run, examine, iterate, driven by frontier models over large-scale compute so thousands of experiments run in parallel instead of sequentially by h...
Use Discovery Loop to turn a recurring-agent idea into an explicit loop contract.
Working implementation, framework, runtime, or repository; signal comes from usable code and ecosystem adoption.
high
README.md
845
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L845
Design
design
Specify a loop contract and operating pattern.
intake
builder
workflow
enabling
implementation
A
ok
https://www.discoveryloop.com/
Discovery Loop — Continuous Exploration
Discovery Loop is building AI systems that automate the experimental loops of science and engineering. Continuous Exploration.
discoveryloop.com
domain-fallback
2026-08-13T14:24:04
ale-0224
Agent Workflow Patterns
agent-workflow-patterns
Docs
📚
Building Effective Agents
https://www.anthropic.com/engineering/building-effective-agents
external
www.anthropic.com
Anthropic's canonical guide to workflows and agents, including evaluator-optimizer and orchestrator-workers patterns.
Anthropic's canonical guide to workflows and agents, including evaluator-optimizer and orchestrator-workers patterns.
Anthropic's canonical guide to workflows and agents, including evaluator-optimizer and orchestrator-workers patterns.
Orchestration and control flow are made explicit and inspectable. Anthropic's canonical guide to workflows and agents, including evaluator-optimizer and orchestrator-workers patterns.
Use Building Effective Agents to turn a recurring-agent idea into an explicit loop contract.
Primary documentation from a platform, SDK, standard, or framework; strong implementation signal.
high
README.md
846
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L846
Design
design
Specify a loop contract and operating pattern.
delegation
builder
workflow
enabling
technical-documentation
A
ok
https://www.anthropic.com/engineering/building-effective-agents
Building Effective AI Agents \ Anthropic
Discover how Anthropic approaches the development of reliable AI agents. Learn about our research on agent capabilities, safety considerations, and technical framework for building trustworthy AI.
Anthropic
domain-fallback
2026-08-13T14:24:04
ale-0225
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Harness Engineering for Language Agents: The Harness Layer as Control, Agency, and Runtime
https://www.preprints.org/manuscript/202603.1756
external
www.preprints.org
Decomposes the harness layer that loops build on into control, agency, and runtime, audits 63 harness works, and proposes a HarnessCard so reported agent gains can be separated from harness effects.
Decomposes the harness layer that loops build on into control, agency, and runtime, audits 63 harness works, and proposes a HarnessCard so reported agent gains can be separated from harness effects.
Decomposes the harness layer that loops build on into control, agency, and runtime, audits 63 harness works, and proposes a HarnessCard so reported agent gains can be separated from harness effects.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Decomposes the harness layer that loops build on into control, agency, and runtime, audits 63 harness works, and proposes a HarnessCard so reported agent gains can be separated from harness effects.
Use Harness Engineering for Language Agents: The Harness Layer as Control, Agency, and Runtime to turn a recurring-agent idea into an explicit loop contract.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
847
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L847
Design
design
Specify a loop contract and operating pattern.
delegation;verification
researcher;evaluator
workflow
enabling
research-paper
A
restricted
https://www.preprints.org/manuscript/202603.1756
Chaoyue He; Xin Zhou; Di Wang; Hong Xu; Wei Liu; Chunyan Miao
2026-04-23
2026
Preprints.org
Preprints.org
10.20944/preprints202603.1756.v2
Version 2; the primary source states that this preprint is not peer-reviewed.
primary-page
2026-08-13T14:24:04
ale-0226
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
How we built our multi-agent research system
https://www.anthropic.com/engineering/multi-agent-research-system
external
www.anthropic.com
Detailed orchestrator-worker system with planning, memory, subagents, citation passes, and iterative research loops.
Detailed orchestrator-worker system with planning, memory, subagents, citation passes, and iterative research loops.
Detailed orchestrator-worker system with planning, memory, subagents, citation passes, and iterative research loops.
Persistent memory is treated as an external runtime artifact. Detailed orchestrator-worker system with planning, memory, subagents, citation passes, and iterative research loops.
Use How we built our multi-agent research system to turn a recurring-agent idea into an explicit loop contract.
Contextual source from www.anthropic.com; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
848
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L848
Design
design
Specify a loop contract and operating pattern.
context;delegation
builder
workflow
enabling
practitioner-analysis
B
ok
https://www.anthropic.com/engineering/multi-agent-research-system
How we built our multi-agent research system \ Anthropic
On the the engineering challenges and lessons learned from building Claude's Research system
Anthropic
domain-fallback
2026-08-13T14:24:04
ale-0227
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Building Effective AI Agents: Architecture Patterns and Implementation Frameworks
https://resources.anthropic.com/hubfs/Building%20Effective%20AI%20Agents-%20Architecture%20Patterns%20and%20Implementation%20Frameworks.pdf
external
resources.anthropic.com
PDF overview of agent architecture patterns, including generator-evaluator loops.
PDF overview of agent architecture patterns, including generator-evaluator loops.
PDF overview of agent architecture patterns, including generator-evaluator loops.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. PDF overview of agent architecture patterns, including generator-evaluator loops.
Use Building Effective AI Agents: Architecture Patterns and Implementation Frameworks to turn a recurring-agent idea into an explicit loop contract.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
849
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L849
Design
design
Specify a loop contract and operating pattern.
delegation;verification
researcher;evaluator
workflow
enabling
research-paper
A
ok
https://resources.anthropic.com/hubfs/Building%20Effective%20AI%20Agents-%20Architecture%20Patterns%20and%20Implementation%20Frameworks.pdf
Anthropic
2025-12-03
2025
Anthropic eBook
Anthropic
Date verified from the primary PDF creation metadata.
pdf-metadata
2026-08-13T14:24:04
ale-0228
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
AI Agent Architectures
https://hld.handbook.academy/curriculum/ai-ml-system-design/ai-agent-architectures/
external
hld.handbook.academy
System-design overview of ReAct, reflection, planning, tool use, memory, and control strategies.
System-design overview of ReAct, reflection, planning, tool use, memory, and control strategies.
System-design overview of ReAct, reflection, planning, tool use, memory, and control strategies.
Persistent memory is treated as an external runtime artifact. System-design overview of ReAct, reflection, planning, tool use, memory, and control strategies.
Use AI Agent Architectures to turn a recurring-agent idea into an explicit loop contract.
Contextual source from hld.handbook.academy; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
850
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L850
Design
design
Specify a loop contract and operating pattern.
workspace;context
builder
workflow
enabling
practitioner-analysis
B
ok
https://hld.handbook.academy/curriculum/ai-ml-system-design/ai-agent-architectures/
AI Agent Architectures (ReAct, Reflection, Planning, Tool Use, Memory) - The HLD Handbook
The canonical patterns for turning an LLM into an agent: ReAct's think-act-observe loop, reflection and self-critique, planner-executor decomposition, tool use and function calling, and how agents manage short- and long-term memory.
The HLD Handbook
html-meta
2026-08-13T14:24:04
ale-0229
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
What Are Agentic Workflows?
https://weaviate.io/blog/what-are-agentic-workflows
external
weaviate.io
Accessible taxonomy of planning, tool use, reflection, and memory patterns.
Accessible taxonomy of planning, tool use, reflection, and memory patterns.
Accessible taxonomy of planning, tool use, reflection, and memory patterns.
Persistent memory is treated as an external runtime artifact. Accessible taxonomy of planning, tool use, reflection, and memory patterns.
Use What Are Agentic Workflows? to turn a recurring-agent idea into an explicit loop contract.
Contextual source from weaviate.io; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
851
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L851
Design
design
Specify a loop contract and operating pattern.
workspace;context
builder
workflow
enabling
practitioner-analysis
B
ok
https://weaviate.io/blog/what-are-agentic-workflows
What Are Agentic Workflows? Patterns, Memory, Use Cases, and Examples | Weaviate
Agentic workflows combine AI agents, tools, and agent memory to create adaptive systems. Learn the core patterns, use cases, and real-world examples.
2025-03-06
2025
weaviate.io
html-meta
2026-08-13T14:24:04
ale-0230
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
Agent Planning & Reflection Patterns
https://learnaivisually.com/tracks/ai-agents/planning-reflection
external
learnaivisually.com
Visual explanation of plan-execute, observe, reflect, retry, and stop patterns.
Visual explanation of plan-execute, observe, reflect, retry, and stop patterns.
Visual explanation of plan-execute, observe, reflect, retry, and stop patterns.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Visual explanation of plan-execute, observe, reflect, retry, and stop patterns.
Use Agent Planning & Reflection Patterns to turn a recurring-agent idea into an explicit loop contract.
Contextual source from learnaivisually.com; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
852
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L852
Design
design
Specify a loop contract and operating pattern.
budget;exit
builder
workflow
enabling
practitioner-analysis
B
ok
https://learnaivisually.com/tracks/ai-agents/planning-reflection
Agent Planning & Reflection Patterns | Learn AI Visually LAV LAV
When agents should plan, retry, pause, or stop. Reasoning budget, ReAct, Reflexion, and termination logic — each tied to a 'when' decision.
Learn AI Visually
html-meta
2026-08-13T14:24:04
ale-0231
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
Agentic Design Patterns
https://addyosmani.com/agents/04-agentic-design-patterns/
external
addyosmani.com
Practical overview of ReAct, reflection, tool use, planning, and how to combine them in real-world agents.
Practical overview of ReAct, reflection, tool use, planning, and how to combine them in real-world agents.
Practical overview of ReAct, reflection, tool use, planning, and how to combine them in real-world agents.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Practical overview of ReAct, reflection, tool use, planning, and how to combine them in real-world agents.
Use Agentic Design Patterns to turn a recurring-agent idea into an explicit loop contract.
Contextual source from addyosmani.com; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
853
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L853
Design
design
Specify a loop contract and operating pattern.
workspace
builder
workflow
enabling
practitioner-analysis
B
ok
https://addyosmani.com/agents/04-agentic-design-patterns/
AddyOsmani.com - Lesson 4: agentic design patterns
Addy Osmani is an engineering and evangelism leader who spent over 14 years at Google leading developer experience across Chrome and, in recent years, AI (Gemini, coding agents, and agentic engineering), most recently as a Director at Google Cloud AI.
Addy Osmani
addyosmani.com
html-meta
2026-08-13T14:24:04
ale-0232
Agent Workflow Patterns
agent-workflow-patterns
Pattern
🔁
12 Factor Agents
https://github.com/humanlayer/12-factor-agents
external
github.com
Operating principles for production agents, including explicit prompts, state ownership, and pause-resume behavior.
Operating principles for production agents, including explicit prompts, state ownership, and pause-resume behavior.
Operating principles for production agents, including explicit prompts, state ownership, and pause-resume behavior.
State persistence is explicit enough for repeated runs and handoff. Operating principles for production agents, including explicit prompts, state ownership, and pause-resume behavior.
Use 12 Factor Agents to turn a recurring-agent idea into an explicit loop contract.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
854
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L854
Design
design
Specify a loop contract and operating pattern.
state
builder
workflow
enabling
operational-pattern
B
ok
https://github.com/humanlayer/12-factor-agents
GitHub - humanlayer/12-factor-agents: What are the principles we can use to build LLM-powered software that is actually good enough to put in the hands of production customers? · GitHub
What are the principles we can use to build LLM-powered software that is actually good enough to put in the hands of production customers? - humanlayer/12-factor-agents
humanlayer/12-factor-agents
GitHub
html-meta
humanlayer/12-factor-agents
2026-08-13T14:24:04
ale-0233
Agent Workflow Patterns
agent-workflow-patterns
Pattern
🔁
Durable Execution for Agentic Workflows
https://arizenai.com/durable-execution/
external
arizenai.com
Explains checkpointing, event-sourced journals, replay, and recovery for long-running agent workflows.
Explains checkpointing, event-sourced journals, replay, and recovery for long-running agent workflows.
Explains checkpointing, event-sourced journals, replay, and recovery for long-running agent workflows.
Durable execution and replay are treated as first-class loop infrastructure. Explains checkpointing, event-sourced journals, replay, and recovery for long-running agent workflows.
Use Durable Execution for Agentic Workflows to turn a recurring-agent idea into an explicit loop contract.
Operational pattern or playbook; signal comes from reusable loop structure and practical transferability.
medium
README.md
855
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L855
Design
design
Specify a loop contract and operating pattern.
state
builder
workflow
enabling
operational-pattern
B
ok
https://arizenai.com/durable-execution/
Durable Execution for Agentic Workflows | Arizen
A while loop is at-most-once across process boundaries. Production agents need exactly-once. The architecture must encode the guarantee.
2026-03-30
2026
Arizen
html-meta
2026-08-13T14:24:04
ale-0234
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Loop Engineering Meets Graph Engineering
gallery/loop-graph-reference.md
local_path
Proposes Loop-Graph, a long-horizon agent framework that couples iterative refinement loops with graph-structured persistent memory; the supplied manuscript screenshot reports evaluation on 9,842 real-world tasks, success gains up to 38.6 percentage points, correctness gains up to 27.4 percentage points, fewer redundan...
Proposes Loop-Graph, a long-horizon agent framework that couples iterative refinement loops with graph-structured persistent memory; the supplied manuscript screenshot reports evaluation on 9,842 real-world tasks, success gains up to 38.6 percentage points, correctness gains up to 27.4 percentage points, fewer redundan...
Proposes Loop-Graph, a long-horizon agent framework that couples iterative refinement loops with graph-structured persistent memory; the supplied manuscript screenshot reports evaluation on 9,842 real-world tasks, success gains up to 38.6 percentage points, correctness gains up to 27.4 percentage points, fewer redundan...
Control flow is represented as an inspectable graph rather than an opaque prompt loop. Proposes Loop-Graph, a long-horizon agent framework that couples iterative refinement loops with graph-structured persistent memory; the supplied manuscript screenshot reports evaluation on 9,842 real-world tasks, success gains up to...
Use Loop Engineering Meets Graph Engineering to turn a recurring-agent idea into an explicit loop contract.
Research source; inspect its method and evaluation before treating results as production evidence.
medium
README.md
856
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L856
2026-07-21
Design
design
Specify a loop contract and operating pattern.
workspace;context;verification;state
researcher;evaluator
workflow
enabling
research-preprint
A
local_ok
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/gallery/loop-graph-reference.md
Loop Engineering Meets Graph Engineering
Lingjiao Chen; Matei Zaharia; Jack Clark; Christopher Re; Chelsea Finn; Ion Stoica
2026
2026
GitHub
GitHub
Author list transcribed from the manuscript screenshot supplied with the gallery reference; kept in code so audit regeneration preserves it.
repository
2026-08-13T14:24:04
ale-0235
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Code as Agent Harness
https://arxiv.org/abs/2605.18747
external
arxiv.org
Organizes agent infrastructure into harness interface, feedback-driven control, and multi-agent scaling for executable, verifiable, stateful systems; maps the harness layer that loops build on.
Organizes agent infrastructure into harness interface, feedback-driven control, and multi-agent scaling for executable, verifiable, stateful systems; maps the harness layer that loops build on.
Organizes agent infrastructure into harness interface, feedback-driven control, and multi-agent scaling for executable, verifiable, stateful systems; maps the harness layer that loops build on.
The work separates roles across agents, verifiers, or orchestration layers. Organizes agent infrastructure into harness interface, feedback-driven control, and multi-agent scaling for executable, verifiable, stateful systems; maps the harness layer that loops build on.
Use Code as Agent Harness to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2605.18747; inspect its method and evaluation before treating results as production evidence.
medium
README.md
857
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L857
Design
design
Specify a loop contract and operating pattern.
delegation;state
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2605.18747
[2605.18747] Code as Agent Harness
Recent large language models (LLMs) have demonstrated strong capabilities in understanding and generating code, from competitive programming to repository-level software engineering. In emerging agentic systems, code is no longer only a target output. It increasingly serves as an operational substrate for agent reasoni...
Xuying Ning; Katherine Tieu; Dongqi Fu; Tianxin Wei; Zihao Li; Yuanchen Bei; Jiaru Zou; Mengting Ai; Zhining Liu; Ting-Wei Li; Lingjie Chen; Yanjun Zhao; Ke Yang; Bingxuan Li; Cheng Qian; Gaotang Li; Xiao Lin; Zhichen Zeng; Ruizhong Qiu; Sirui Chen; Yifan Sun; Xiyuan Yang; Ruida Wang; Rui Pan; Chenyuan Yang; Dylan Zhan...
2026-05-18
2026
arXiv
arXiv
GitHub: https://github.com/YennNing/Awesome-Code-as-Agent-Harness-Papers
cs.CL
arxiv-api
2605.18747
2026-08-13T14:24:04
ale-0236
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Agentic Agile-V: From Vibe Coding to Verified Engineering
https://arxiv.org/abs/2605.20456
external
arxiv.org
Proposes a task-level SCOPE-V loop (Specify, Constrain, Orchestrate, Prove, Evolve, Verify) with human approval gates, arguing agentic coding needs process control and independent verification, not better prompts.
Proposes a task-level SCOPE-V loop (Specify, Constrain, Orchestrate, Prove, Evolve, Verify) with human approval gates, arguing agentic coding needs process control and independent verification, not better prompts.
Proposes a task-level SCOPE-V loop (Specify, Constrain, Orchestrate, Prove, Evolve, Verify) with human approval gates, arguing agentic coding needs process control and independent verification, not better prompts.
Verification is promoted from a final check to a loop-control signal. Proposes a task-level SCOPE-V loop (Specify, Constrain, Orchestrate, Prove, Evolve, Verify) with human approval gates, arguing agentic coding needs process control and independent verification, not better prompts.
Use Agentic Agile-V: From Vibe Coding to Verified Engineering to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2605.20456; inspect its method and evaluation before treating results as production evidence.
medium
README.md
858
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L858
Design
design
Specify a loop contract and operating pattern.
delegation;verification;escalation
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2605.20456
[2605.20456] Agentic Agile-V: From Vibe Coding to Verified Engineering in Software and Hardware Development
Agentic AI coding systems can inspect repositories, plan implementation steps, edit files, call tools, run tests, and submit pull requests. These capabilities make software and hardware development faster in some settings, but current evidence does not support the simple claim that autonomous code generation automatica...
Christopher Koch
2026-05-19
2026
arXiv
arXiv
7 pages, 1 figure
cs.SE
arxiv-api
2605.20456
2026-08-13T14:24:04
ale-0237
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Agentic Software Engineering: Foundational Pillars and a Research Roadmap
https://arxiv.org/abs/2509.06216
external
arxiv.org
Splits agentic SE into an Agent Command Environment for human orchestration and an Agent Execution Environment for agent task execution, a research roadmap for the layers recurring loops run inside.
Splits agentic SE into an Agent Command Environment for human orchestration and an Agent Execution Environment for agent task execution, a research roadmap for the layers recurring loops run inside.
Splits agentic SE into an Agent Command Environment for human orchestration and an Agent Execution Environment for agent task execution, a research roadmap for the layers recurring loops run inside.
Orchestration and control flow are made explicit and inspectable. Splits agentic SE into an Agent Command Environment for human orchestration and an Agent Execution Environment for agent task execution, a research roadmap for the layers recurring loops run inside.
Use Agentic Software Engineering: Foundational Pillars and a Research Roadmap to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2509.06216; inspect its method and evaluation before treating results as production evidence.
medium
README.md
859
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L859
Design
design
Specify a loop contract and operating pattern.
delegation;escalation
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2509.06216
[2509.06216] Agentic Software Engineering: Foundational Pillars and a Research Roadmap
Agentic Software Engineering (SE 3.0) represents a new era where intelligent agents are tasked not with simple code generation, but with achieving complex, goal-oriented SE objectives. To harness these new capabilities while ensuring trustworthiness, we must recognize a fundamental duality within the SE field in the Ag...
Ahmed E. Hassan; Hao Li; Dayi Lin; Bram Adams; Tse-Hsun Chen; Yutaro Kashiwa; Dong Qiu
2025-09-07
2025
arXiv
arXiv
cs.SE
arxiv-api
2509.06216
2026-08-13T14:24:04
ale-0238
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
The Art of Loop Engineering
https://www.langchain.com/blog/the-art-of-loop-engineering
external
www.langchain.com
LangChain's account of four stacked loops around agents (core execution, rubric-based verification, event-driven triggers, and trace-driven self-improvement) using a documentation-writing agent as the running example.
LangChain's account of four stacked loops around agents (core execution, rubric-based verification, event-driven triggers, and trace-driven self-improvement) using a documentation-writing agent as the running example.
LangChain's account of four stacked loops around agents (core execution, rubric-based verification, event-driven triggers, and trace-driven self-improvement) using a documentation-writing agent as the running example.
Verification is promoted from a final check to a loop-control signal. LangChain's account of four stacked loops around agents (core execution, rubric-based verification, event-driven triggers, and trace-driven self-improvement) using a documentation-writing agent as the running example.
Use The Art of Loop Engineering to turn a recurring-agent idea into an explicit loop contract.
Contextual source from www.langchain.com; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
860
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L860
Design
design
Specify a loop contract and operating pattern.
trigger;verification
builder
workflow
enabling
practitioner-analysis
B
ok
https://www.langchain.com/blog/the-art-of-loop-engineering
The Art of Loop Engineering
Agents automate real-world work, but reliable performance requires more than a good model, it requires a carefully designed harness built for specific tasks. This post explores the core agent loop, how stacking and extending loops builds more effective agents, and how to instrument each level with LangChain primitives.
LangChain
domain-fallback
2026-08-13T14:24:04
ale-0239
Agent Workflow Patterns
agent-workflow-patterns
Tool
🧰
Loopy
https://github.com/Forward-Future/loopy
external
github.com
Library of reusable AI-agent loops with verification checks and stopping conditions, plus an installable skill for finding, adapting, and designing repeatable agent workflows.
Library of reusable AI-agent loops with verification checks and stopping conditions, plus an installable skill for finding, adapting, and designing repeatable agent workflows.
Library of reusable AI-agent loops with verification checks and stopping conditions, plus an installable skill for finding, adapting, and designing repeatable agent workflows.
Verification is promoted from a final check to a loop-control signal. Library of reusable AI-agent loops with verification checks and stopping conditions, plus an installable skill for finding, adapting, and designing repeatable agent workflows.
Use Loopy to turn a recurring-agent idea into an explicit loop contract.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
861
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L861
Design
design
Specify a loop contract and operating pattern.
verification;exit
builder
workflow
enabling
source-implementation
A
ok
https://github.com/Forward-Future/loopy
GitHub - Forward-Future/loopy: A library of practical AI-agent loops and an installable skill for finding, adapting, and designing repeatable agent workflows. · GitHub
A library of practical AI-agent loops and an installable skill for finding, adapting, and designing repeatable agent workflows. - Forward-Future/loopy
Forward-Future/loopy
GitHub
html-meta
Forward-Future/loopy
2026-08-13T14:24:04
ale-0240
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
The Factory Model: How Coding Agents Changed Software Engineering
https://addyosmani.com/blog/factory-model/
external
addyosmani.com
Addy Osmani's February 2026 essay recasting engineers as operators of parallel agent workstreams, with agents running "thirty minutes, an hour, several hours and increasingly days" and verification, not generation, as the new bottleneck; the fleet-level framing that precedes his June Loop Engineering essay.
Addy Osmani's February 2026 essay recasting engineers as operators of parallel agent workstreams, with agents running "thirty minutes, an hour, several hours and increasingly days" and verification, not generation, as the new bottleneck; the fleet-level framing that precedes his June Loop Engineering essay.
Addy Osmani's February 2026 essay recasting engineers as operators of parallel agent workstreams, with agents running "thirty minutes, an hour, several hours and increasingly days" and verification, not generation, as the new bottleneck; the fleet-level framing that precedes his June Loop Engineering essay.
Verification is promoted from a final check to a loop-control signal. Addy Osmani's February 2026 essay recasting engineers as operators of parallel agent workstreams, with agents running "thirty minutes, an hour, several hours and increasingly days" and verification, not generation, as the new bottleneck; the fleet-le...
Use The Factory Model: How Coding Agents Changed Software Engineering to turn a recurring-agent idea into an explicit loop contract.
Contextual source from addyosmani.com; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
862
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L862
Design
design
Specify a loop contract and operating pattern.
verification
builder
workflow
enabling
practitioner-analysis
B
ok
https://addyosmani.com/blog/factory-model/
AddyOsmani.com - The Factory Model: How Coding Agents Changed Software Engineering
Software engineering is not about writing code anymore. It is about building the factory that builds your software.
Addy Osmani
addyosmani.com
html-meta
2026-08-13T14:24:04
ale-0241
Agent Workflow Patterns
agent-workflow-patterns
Docs
📚
2026 Agentic Coding Trends Report
https://resources.anthropic.com/2026-agentic-coding-trends-report
external
resources.anthropic.com
Anthropic's official 2026 trends report on the shift from single coding assistants to coordinated agent teams running autonomously for hours or days, with case studies from Rakuten, TELUS, and Zapier (landing page is registration-gated; the direct PDF is public).
Anthropic's official 2026 trends report on the shift from single coding assistants to coordinated agent teams running autonomously for hours or days, with case studies from Rakuten, TELUS, and Zapier (landing page is registration-gated; the direct PDF is public).
Anthropic's official 2026 trends report on the shift from single coding assistants to coordinated agent teams running autonomously for hours or days, with case studies from Rakuten, TELUS, and Zapier (landing page is registration-gated; the direct PDF is public).
Primary-source operational guidance rather than commentary. Anthropic's official 2026 trends report on the shift from single coding assistants to coordinated agent teams running autonomously for hours or days, with case studies from Rakuten, TELUS, and Zapier (landing page is registration-gated; the direct PDF is publi...
Use 2026 Agentic Coding Trends Report to turn a recurring-agent idea into an explicit loop contract.
Primary documentation from a platform, SDK, standard, or framework; strong implementation signal.
high
README.md
863
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L863
Design
design
Specify a loop contract and operating pattern.
delegation;verification
builder
workflow
enabling
technical-documentation
A
ok
https://resources.anthropic.com/2026-agentic-coding-trends-report
2026 Agentic Coding Trends Report
How coding agents are transforming software development - and what it means for engineering teams in 2026. Insights on multi-agent systems, human-AI collaboration, and scaling agentic coding across organizations. Includes case studies from Rakuten, TELUS, Zapier, and more.
2026
Anthropic
url-date
2026-08-13T14:24:04
ale-0242
Agent Workflow Patterns
agent-workflow-patterns
Tool
🧰
HomeRail
https://github.com/xiaotianfotos/homerail
external
github.com
TypeScript runtime that turns one-off agent chats into auditable, reusable DAG workflows on self-hosted hardware, with a DAG engine, CLI, and voice front end.
TypeScript runtime that turns one-off agent chats into auditable, reusable DAG workflows on self-hosted hardware, with a DAG engine, CLI, and voice front end.
TypeScript runtime that turns one-off agent chats into auditable, reusable DAG workflows on self-hosted hardware, with a DAG engine, CLI, and voice front end.
Control flow is represented as an inspectable graph rather than an opaque prompt loop. TypeScript runtime that turns one-off agent chats into auditable, reusable DAG workflows on self-hosted hardware, with a DAG engine, CLI, and voice front end.
Use HomeRail to turn a recurring-agent idea into an explicit loop contract.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
864
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L864
Design
design
Specify a loop contract and operating pattern.
delegation;verification
builder
workflow
enabling
source-implementation
A
ok
https://github.com/xiaotianfotos/homerail
GitHub - xiaotianfotos/homerail: Voice-first local agent orchestration runtime for auditable DAG workflows. · GitHub
Voice-first local agent orchestration runtime for auditable DAG workflows. - xiaotianfotos/homerail
xiaotianfotos/homerail
GitHub
html-meta
xiaotianfotos/homerail
2026-08-13T14:24:04
ale-0243
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
Old and New Apps, via Modern Coding Agents
https://terrytao.wordpress.com/2026/07/11/old-and-new-apps-via-modern-coding-agents/
external
terrytao.wordpress.com
Terence Tao's July 11, 2026 account of porting roughly two dozen Java 1.0 applets to JavaScript and resurrecting long-abandoned projects through iterative agent sessions, concluding that domain expertise remains the human verification layer: high-level design decisions stay with the author while implementation is autom...
Terence Tao's July 11, 2026 account of porting roughly two dozen Java 1.0 applets to JavaScript and resurrecting long-abandoned projects through iterative agent sessions, concluding that domain expertise remains the human verification layer: high-level design decisions stay with the author while implementation is autom...
Terence Tao's July 11, 2026 account of porting roughly two dozen Java 1.0 applets to JavaScript and resurrecting long-abandoned projects through iterative agent sessions, concluding that domain expertise remains the human verification layer: high-level design decisions stay with the author while implementation is autom...
Verification is promoted from a final check to a loop-control signal. Terence Tao's July 11, 2026 account of porting roughly two dozen Java 1.0 applets to JavaScript and resurrecting long-abandoned projects through iterative agent sessions, concluding that domain expertise remains the human verification layer: high-lev...
Use Old and New Apps, via Modern Coding Agents to turn a recurring-agent idea into an explicit loop contract.
Contextual source from terrytao.wordpress.com; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
865
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L865
Design
design
Specify a loop contract and operating pattern.
verification;escalation
builder
workflow
enabling
practitioner-analysis
B
ok
https://terrytao.wordpress.com/2026/07/11/old-and-new-apps-via-modern-coding-agents/
Old and new apps, via modern coding agents | What's new
I have been interested in machine-assisted ways to do and teach mathematics from as far back as 1999, when I started coding several applets in Java 1.0, both for my complex analysis and linear alge…
2026-07-11
2026
What's new
html-meta
2026-08-13T14:24:04
ale-0244
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable
https://arxiv.org/abs/2607.13285
external
arxiv.org
Introduces a behavior-centered source map and behavior-guided program decomposition for evolving harnesses, improving behavior localization and edit planning on two agent harnesses while reducing planner token use.
Introduces a behavior-centered source map and behavior-guided program decomposition for evolving harnesses, improving behavior localization and edit planning on two agent harnesses while reducing planner token use.
Introduces a behavior-centered source map and behavior-guided program decomposition for evolving harnesses, improving behavior localization and edit planning on two agent harnesses while reducing planner token use.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Introduces a behavior-centered source map and behavior-guided program decomposition for evolving harnesses, improving behavior localization and edit planning on two agent harnesses while reducing planner token use.
Use Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.13285; inspect its method and evaluation before treating results as production evidence.
medium
README.md
866
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L866
2026-07-17
Design
design
Specify a loop contract and operating pattern.
budget
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.13285
[2607.13285] Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable
The capability of a modern AI agent depends not only on its foundation model but also on its harness, which constructs prompts, manages state, invokes tools, and coordinates execution. As models, APIs, environments, and requirements evolve, the harness must be continually modified. Before such a change can be made, a d...
Ruhan Wang; Yucheng Shi; Zongxia Li; Zhongzhi Li; Yue Yu; Junyao Yang; Kishan Panaganti; Haitao Mi; Dongruo Zhou; Leoweiliang
2026-07-14
2026
arXiv
arXiv
29 pages, 6 figures. Project page: https://ruhan-wang.github.io/Harness-Handbook/
cs.AI
arxiv-api
2607.13285
2026-08-13T14:24:04
ale-0245
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
MemoHarness: Agent Harnesses That Learn from Experience
https://arxiv.org/abs/2607.14159
external
arxiv.org
Lets an agent adapt six harness dimensions per case using a dual-layer experience memory; results improve over fixed harnesses, while broad robustness and the contribution of each adaptive component remain open questions.
Lets an agent adapt six harness dimensions per case using a dual-layer experience memory; results improve over fixed harnesses, while broad robustness and the contribution of each adaptive component remain open questions.
Lets an agent adapt six harness dimensions per case using a dual-layer experience memory; results improve over fixed harnesses, while broad robustness and the contribution of each adaptive component remain open questions.
Persistent memory is treated as an external runtime artifact. Lets an agent adapt six harness dimensions per case using a dual-layer experience memory; results improve over fixed harnesses, while broad robustness and the contribution of each adaptive component remain open questions.
Use MemoHarness: Agent Harnesses That Learn from Experience to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.14159; inspect its method and evaluation before treating results as production evidence.
medium
README.md
867
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L867
2026-07-17
Design
design
Specify a loop contract and operating pattern.
context
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.14159
[2607.14159] MemoHarness: Agent Harnesses That Learn from Experience
An agent harness is the external control layer that turns a base LLM into an executable agent by managing context, tools, orchestration, memory, decoding, and output handling. While harness design strongly affects agent behavior, most automatic improvement methods optimize narrower artifacts such as prompts, pipelines,...
Yue Huang; Wenjie Wang; Han Bao; Yuchen Ma; Xiaonan Luo; Yi Nian; Haomin Zhuang; Zheyuan Liu; Yue Zhao; Xiangliang Zhang
2026-07-14
2026
arXiv
arXiv
cs.AI
arxiv-api
2607.14159
2026-08-13T14:24:04
ale-0246
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Recursive Harness Self-Improvement
https://arxiv.org/abs/2607.15524
external
arxiv.org
Represents an agent loop as a prompt-level harness specification and iteratively revises it from pairwise feedback over its own history; across 30 synthetic machine-learning research tasks, a few revisions lift low-reasoning agents above the corresponding maximum-reasoning setting while cutting inference cost by up to ...
Represents an agent loop as a prompt-level harness specification and iteratively revises it from pairwise feedback over its own history; across 30 synthetic machine-learning research tasks, a few revisions lift low-reasoning agents above the corresponding maximum-reasoning setting while cutting inference cost by up to ...
Represents an agent loop as a prompt-level harness specification and iteratively revises it from pairwise feedback over its own history; across 30 synthetic machine-learning research tasks, a few revisions lift low-reasoning agents above the corresponding maximum-reasoning setting while cutting inference cost by up to ...
Context is managed as durable loop state rather than a single prompt payload. Represents an agent loop as a prompt-level harness specification and iteratively revises it from pairwise feedback over its own history; across 30 synthetic machine-learning research tasks, a few revisions lift low-reasoning agents above the ...
Use Recursive Harness Self-Improvement to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.15524; inspect its method and evaluation before treating results as production evidence.
medium
README.md
868
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L868
2026-07-20
Design
design
Specify a loop contract and operating pattern.
context;budget
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.15524
[2607.15524] Recursive Harness Self-Improvement
Under model--harness co-evolution, harnesses are not merely inference-time scaffolds but data-generating components whose execution traces can shape future foundation models. This motivates harness-in-the-loop learning: optimizing harnesses for both immediate agent performance and the quality of traces used for future ...
Hyunin Lee; Jinglue Xu; Jeffrey Seely; Donghyun Lee; Matei Zaharia; Yujin Tang
2026-07-17
2026
arXiv
arXiv
This work addresses the first half of the model-harness coevolution loop
cs.LG
arxiv-api
2607.15524
2026-08-13T14:24:04
ale-0247
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
SkillCorpus: Consolidating and Evaluating the Open Skill Ecosystem for Real-World LLM Agents
https://arxiv.org/abs/2607.15557
external
arxiv.org
Filters roughly 821,000 crawled agent skills into a 96,401-item corpus with a 16-class taxonomy and utility, robustness, and safety facets; evaluations across three benchmarks, two harnesses, and two open models report gains up to 7.5 percentage points while exposing both coverage and harness boundaries. The announced ...
Filters roughly 821,000 crawled agent skills into a 96,401-item corpus with a 16-class taxonomy and utility, robustness, and safety facets; evaluations across three benchmarks, two harnesses, and two open models report gains up to 7.5 percentage points while exposing both coverage and harness boundaries. The announced ...
Filters roughly 821,000 crawled agent skills into a 96,401-item corpus with a 16-class taxonomy and utility, robustness, and safety facets; evaluations across three benchmarks, two harnesses, and two open models report gains up to 7.5 percentage points while exposing both coverage and harness boundaries. The announced ...
Packages the evidence as queryable CSV and JSONL rather than only a rendered page. Filters roughly 821,000 crawled agent skills into a 96,401-item corpus with a 16-class taxonomy and utility, robustness, and safety facets; evaluations across three benchmarks, two harnesses, and two open models report gains up to 7.5 pe...
Use SkillCorpus: Consolidating and Evaluating the Open Skill Ecosystem for Real-World LLM Agents to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.15557; inspect its method and evaluation before treating results as production evidence.
medium
README.md
869
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L869
2026-07-20
Design
design
Specify a loop contract and operating pattern.
verification
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.15557
[2607.15557] SkillCorpus: Consolidating and Evaluating the Open Skill Ecosystem for Real-World LLM Agents
Agent skills, SKILL files that package reusable procedural knowledge for an LLM agent, are a popular mechanism for extending agent capabilities. Public repositories now host them in large and growing numbers, yet these artifacts are fragmented, redundant, and uneven in quality, and their value in practice is unclear. A...
Yanze Wang; Pengfei Yao; Tianyi Sun; Chuanrui Hu; Yan Xiao; Xiaotian Luo; Yunyun Han; Yifan Chen; Jun Sun; Yafeng Deng
2026-07-17
2026
arXiv
arXiv
cs.CL
arxiv-api
2607.15557
2026-08-13T14:24:04
ale-0248
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Behavioral Controllability of Agentic Models for Information Extraction: From Fixed Workflows to Reflective Agents
https://arxiv.org/abs/2607.15715
external
arxiv.org
Compares a fixed extraction workflow with reflective variants under one evaluation harness, measuring tool execution, retries, reflection, memory use, runtime, and recovery so agentic mechanisms can be judged by observable process changes rather than output scores alone.
Compares a fixed extraction workflow with reflective variants under one evaluation harness, measuring tool execution, retries, reflection, memory use, runtime, and recovery so agentic mechanisms can be judged by observable process changes rather than output scores alone.
Compares a fixed extraction workflow with reflective variants under one evaluation harness, measuring tool execution, retries, reflection, memory use, runtime, and recovery so agentic mechanisms can be judged by observable process changes rather than output scores alone.
Evaluation data is used as the feedback signal for improving loop behavior. Compares a fixed extraction workflow with reflective variants under one evaluation harness, measuring tool execution, retries, reflection, memory use, runtime, and recovery so agentic mechanisms can be judged by observable process changes rathe...
Use Behavioral Controllability of Agentic Models for Information Extraction: From Fixed Workflows to Reflective Agents to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.15715; inspect its method and evaluation before treating results as production evidence.
medium
README.md
870
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L870
2026-07-20
Design
design
Specify a loop contract and operating pattern.
workspace;context;verification;budget
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.15715
[2607.15715] Behavioral Controllability of Agentic Models for Information Extraction: From Fixed Workflows to Reflective Agents
Large language model (LLM) agents are increasingly used for complex information-extraction tasks, yet it remains unclear whether agentic components such as reflection and memory lead to observable and controllable improvements over fixed LLM workflows. We study this question through conference-paper dataset extraction,...
Lujia Zhang; Xingzhou Chen; Hongwei Feng
2026-07-17
2026
arXiv
arXiv
cs.AI
arxiv-api
2607.15715
2026-08-13T14:24:04
ale-0249
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Understanding Agent-Reactive Bugs at the Model-Harness Boundary: An Empirical Study of LLM Agent Issue Reports
https://arxiv.org/abs/2607.15684
external
arxiv.org
Manually analyzes 255 issue reports from Codex, Gemini CLI, LangChain, and CrewAI to classify bugs triggered by an interaction between model output and harness behavior; silent failures, weak test oracles, and stochastic reproduction motivate dedicated replay and fault-localization support.
Manually analyzes 255 issue reports from Codex, Gemini CLI, LangChain, and CrewAI to classify bugs triggered by an interaction between model output and harness behavior; silent failures, weak test oracles, and stochastic reproduction motivate dedicated replay and fault-localization support.
Manually analyzes 255 issue reports from Codex, Gemini CLI, LangChain, and CrewAI to classify bugs triggered by an interaction between model output and harness behavior; silent failures, weak test oracles, and stochastic reproduction motivate dedicated replay and fault-localization support.
Durable execution and replay are treated as first-class loop infrastructure. Manually analyzes 255 issue reports from Codex, Gemini CLI, LangChain, and CrewAI to classify bugs triggered by an interaction between model output and harness behavior; silent failures, weak test oracles, and stochastic reproduction motivate ...
Use Understanding Agent-Reactive Bugs at the Model-Harness Boundary: An Empirical Study of LLM Agent Issue Reports to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.15684; inspect its method and evaluation before treating results as production evidence.
medium
README.md
871
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L871
2026-07-20
Design
design
Specify a loop contract and operating pattern.
trigger;intake;verification;state
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.15684
[2607.15684] Understanding Agent-Reactive Bugs at the Model-Harness Boundary: An Empirical Study of LLM Agent Issue Reports
LLM agents span command-line interfaces (e.g., Codex) and agent frameworks (e.g., LangChain), integrating backend LLMs with harness code that parses model outputs, controls agent loops, and manages context. Both the harness and LLM-generated responses jointly shape an agent's execution. This architecture gives rise to ...
Jingyi Chen; Songqiang Chen; Hengcheng Zhu; Jialun Cao; Jiasi Shen; Shing-Chi Cheung
2026-07-17
2026
arXiv
arXiv
cs.SE
arxiv-api
2607.15684
2026-08-13T14:24:04
ale-0250
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
SciForge: An AI-Native, Multimodal Workbench for Scientific Discovery
https://arxiv.org/abs/2607.16038
external
arxiv.org
Organizes long-running scientific work around goal-scoped review gates, domain translators, an agent runtime and workflow engine, and an evidence DAG that links claims to provenance; eight end-to-end cases include multi-day research sprints, while deeper team collaboration remains planned work.
Organizes long-running scientific work around goal-scoped review gates, domain translators, an agent runtime and workflow engine, and an evidence DAG that links claims to provenance; eight end-to-end cases include multi-day research sprints, while deeper team collaboration remains planned work.
Organizes long-running scientific work around goal-scoped review gates, domain translators, an agent runtime and workflow engine, and an evidence DAG that links claims to provenance; eight end-to-end cases include multi-day research sprints, while deeper team collaboration remains planned work.
Control flow is represented as an inspectable graph rather than an opaque prompt loop. Organizes long-running scientific work around goal-scoped review gates, domain translators, an agent runtime and workflow engine, and an evidence DAG that links claims to provenance; eight end-to-end cases include multi-day research ...
Use SciForge: An AI-Native, Multimodal Workbench for Scientific Discovery to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.16038; inspect its method and evaluation before treating results as production evidence.
medium
README.md
872
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L872
2026-07-20
Design
design
Specify a loop contract and operating pattern.
objective;intake
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.16038
[2607.16038] SciForge: An AI-Native, Multimodal Workbench for Scientific Discovery
Scientific work increasingly spans heterogeneous artifacts -- papers, code, datasets, scientific file formats, model outputs, figures, manuscripts, and team decisions -- yet general-purpose AI assistants rarely preserve these objects as a coherent, auditable research state. We present SciForge, a multimodal research-na...
SciForge Team; Zhangyang Gao; Minghao Fang; Yifei Liu; Hanhui Yang; Xinyu Gu; Shixiang Tang; Siqi Sun; Lei Bai; Cheng Tan; Mengdi Liu; Hao Wu; Shuizhou Chen
2026-07-17
2026
arXiv
arXiv
cs.AI
arxiv-api
2607.16038
2026-08-13T14:24:04
ale-0251
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
Coding Agents 2.0: Interface, Inference, and Verification
https://www.gradient.com/blog/posts/coding-agents-2/
external
www.gradient.com
Argues the coding-agent bottleneck has shifted from generation to three infrastructure problems: interfaces for parallel agent fleets, a coding-optimized inference stack with prefix caching, and verification layers built on formal proofs and trace diffing.
Argues the coding-agent bottleneck has shifted from generation to three infrastructure problems: interfaces for parallel agent fleets, a coding-optimized inference stack with prefix caching, and verification layers built on formal proofs and trace diffing.
Argues the coding-agent bottleneck has shifted from generation to three infrastructure problems: interfaces for parallel agent fleets, a coding-optimized inference stack with prefix caching, and verification layers built on formal proofs and trace diffing.
Verification is promoted from a final check to a loop-control signal. Argues the coding-agent bottleneck has shifted from generation to three infrastructure problems: interfaces for parallel agent fleets, a coding-optimized inference stack with prefix caching, and verification layers built on formal proofs and trace di...
Use Coding Agents 2.0: Interface, Inference, and Verification to turn a recurring-agent idea into an explicit loop contract.
Contextual source from www.gradient.com; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
873
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L873
2026-07-22
Design
design
Specify a loop contract and operating pattern.
verification
builder
workflow
enabling
practitioner-analysis
B
ok
https://www.gradient.com/blog/posts/coding-agents-2/
Coding Agents 2.0: Interface, Inference, and Verification | Gradient Ventures The coding agent stack — Market map
In under a year, writing software became the biggest thing people do with large language models. The revenue followed the tokens: Cursor reached about $4B in annualized revenue in under four years and recently <a href="https://techcrunch.com/2026/04/22/how-spacex-preempted-a-2b-fundraise-with-a-60b-buyout-offer/">agree...
Gradient Ventures
html-meta
2026-08-13T14:24:04
ale-0252
Agent Workflow Patterns
agent-workflow-patterns
Blog
📝
Towards a Harness That Can Do Anything
https://eardatasci.github.io/c/ambiance/index.html
external
eardatasci.github.io
Essay on designing a general-purpose agent harness, examining what capability boundaries, tool surfaces, and verification hooks a do-anything harness actually requires.
Essay on designing a general-purpose agent harness, examining what capability boundaries, tool surfaces, and verification hooks a do-anything harness actually requires.
Essay on designing a general-purpose agent harness, examining what capability boundaries, tool surfaces, and verification hooks a do-anything harness actually requires.
Verification is promoted from a final check to a loop-control signal. Essay on designing a general-purpose agent harness, examining what capability boundaries, tool surfaces, and verification hooks a do-anything harness actually requires.
Use Towards a Harness That Can Do Anything to turn a recurring-agent idea into an explicit loop contract.
Contextual source from eardatasci.github.io; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
874
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L874
2026-07-22
Design
design
Specify a loop contract and operating pattern.
workspace;verification
builder
workflow
enabling
practitioner-analysis
B
ok
https://eardatasci.github.io/c/ambiance/index.html
arda tasci
eardatasci.github.io
domain-fallback
2026-08-13T14:24:04
ale-0253
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
NVIDIA-labs OO Agents: Native Python Object-Oriented Agents
https://arxiv.org/abs/2607.20709
external
arxiv.org
NVIDIA's model-agnostic framework (NOOA) where an agent is a Python object: methods are actions, fields are state, docstrings are prompts, and type annotations are contracts; a method body of '...' is completed at runtime by an LLM-driven agent loop while normal bodies stay deterministic, letting agent behavior be test...
NVIDIA's model-agnostic framework (NOOA) where an agent is a Python object: methods are actions, fields are state, docstrings are prompts, and type annotations are contracts; a method body of '...' is completed at runtime by an LLM-driven agent loop while normal bodies stay deterministic, letting agent behavior be test...
NVIDIA's model-agnostic framework (NOOA) where an agent is a Python object: methods are actions, fields are state, docstrings are prompts, and type annotations are contracts; a method body of '...' is completed at runtime by an LLM-driven agent loop while normal bodies stay deterministic, letting agent behavior be test...
State persistence is explicit enough for repeated runs and handoff. NVIDIA's model-agnostic framework (NOOA) where an agent is a Python object: methods are actions, fields are state, docstrings are prompts, and type annotations are contracts; a method body of '...' is completed at runtime by an LLM-driven agent loop wh...
Use NVIDIA-labs OO Agents: Native Python Object-Oriented Agents to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.20709; inspect its method and evaluation before treating results as production evidence.
medium
README.md
875
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L875
2026-07-24
Design
design
Specify a loop contract and operating pattern.
verification;state
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.20709
[2607.20709] NVIDIA-labs OO Agents: Native Python Object-Oriented Agents
Traditional agent development is split across prompt templates, tool schemas, callback code, and workflow graphs. We present NVIDIA Object-Oriented Agents (NOOA), a model-agnostic Python framework for building reliable AI agents. NOOA takes a simpler approach: an agent is a Python object. Its methods are the actions th...
Paul Furgale; Severin Klingler; James Nolan; Matt Staats; Gaia Di Lorenzo; Elisa Martinez Abad; Christian Schüller; Razvan Dinu; Alessio Devoto; Pascal Berard; Gal Kaplun; Elad Sarafian; Riccardo Roveri; Leon Derczynski; Ricardo Silveira Cabral
2026-07-22
2026
arXiv
arXiv
cs.AI
arxiv-api
2607.20709
2026-08-13T14:24:04
ale-0254
Agent Workflow Patterns
agent-workflow-patterns
Tool
🧰
deer-workflow
https://github.com/deerwork-ai/deer-workflow
external
github.com
Code-first runtime that keeps orchestration and control flow in TypeScript while delegating only the semantic work to replaceable agent runtimes (Codex CLI and Claude supported out of the box). Emits an observable JSONL event stream for downstream automation and ships an interactive TUI for watching execution.
Code-first runtime that keeps orchestration and control flow in TypeScript while delegating only the semantic work to replaceable agent runtimes (Codex CLI and Claude supported out of the box). Emits an observable JSONL event stream for downstream automation and ships an interactive TUI for watching execution.
Code-first runtime that keeps orchestration and control flow in TypeScript while delegating only the semantic work to replaceable agent runtimes (Codex CLI and Claude supported out of the box). Emits an observable JSONL event stream for downstream automation and ships an interactive TUI for watching execution.
Orchestration and control flow are made explicit and inspectable. Code-first runtime that keeps orchestration and control flow in TypeScript while delegating only the semantic work to replaceable agent runtimes (Codex CLI and Claude supported out of the box). Emits an observable JSONL event stream for downstream automa...
Use deer-workflow to turn a recurring-agent idea into an explicit loop contract.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
876
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L876
2026-07-28
Design
design
Specify a loop contract and operating pattern.
delegation
builder
workflow
enabling
source-implementation
A
ok
https://github.com/deerwork-ai/deer-workflow
GitHub - deerwork-ai/deer-workflow: An open-source graph engineering runtime that keeps orchestration in TypeScript and delegates semantic work to replaceable Agent runtimes. · GitHub
An open-source graph engineering runtime that keeps orchestration in TypeScript and delegates semantic work to replaceable Agent runtimes. - deerwork-ai/deer-workflow
deerwork-ai/deer-workflow
GitHub
html-meta
deerwork-ai/deer-workflow
2026-08-13T14:24:04
ale-0255
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Living-Harness Is an Interactive-Agent Evolver
https://arxiv.org/abs/2607.26598
external
arxiv.org
Converts each completed trajectory and its evaluator signals into posterior evidence for bounded harness updates, so the scaffold itself evolves via episodic memory and state graphs instead of staying static. Directly operationalizes the self-improving-loop premise: the harness is the learned artifact, not the weights.
Converts each completed trajectory and its evaluator signals into posterior evidence for bounded harness updates, so the scaffold itself evolves via episodic memory and state graphs instead of staying static. Directly operationalizes the self-improving-loop premise: the harness is the learned artifact, not the weights.
Converts each completed trajectory and its evaluator signals into posterior evidence for bounded harness updates, so the scaffold itself evolves via episodic memory and state graphs instead of staying static. Directly operationalizes the self-improving-loop premise: the harness is the learned artifact, not the weights.
Control flow is represented as an inspectable graph rather than an opaque prompt loop. Converts each completed trajectory and its evaluator signals into posterior evidence for bounded harness updates, so the scaffold itself evolves via episodic memory and state graphs instead of staying static. Directly operationalizes...
Use Living-Harness Is an Interactive-Agent Evolver to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.26598; inspect its method and evaluation before treating results as production evidence.
medium
README.md
877
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L877
2026-07-30
Design
design
Specify a loop contract and operating pattern.
context;state
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.26598
[2607.26598] Living-Harness Is an Interactive-Agent Evolver
Large language model (LLM) agents may recover from a failure within an episode or after a retry, yet the same execution failure can recur in later tasks because post-episode feedback rarely revises the persistent harness that guides future interactions. Static harnesses improve reliability through fixed tools, context,...
Yuetian Du; Yucheng Wang; He Xu; Jiexu Xu; Shanwen Tan; Bing Zhao; Boyu Yang; Zhijie Xu; Ming Kong; Hu Wei; Jie Liu; Qiang Zhu
2026-07-29
2026
arXiv
arXiv
cs.MA
arxiv-api
2607.26598
2026-08-13T14:24:04
ale-0256
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
CHILL-Harness: Counterfactual Harness Learning for Efficient Reasoning in Long-Horizon Agents
https://arxiv.org/abs/2607.25825
external
arxiv.org
Uses counterfactual causal learning to let an agent's harness adapt orchestration per task and environment, cutting long-horizon compute while holding success rate. One of the first papers to treat harness configuration as a learnable causal policy rather than a hand-tuned constant.
Uses counterfactual causal learning to let an agent's harness adapt orchestration per task and environment, cutting long-horizon compute while holding success rate. One of the first papers to treat harness configuration as a learnable causal policy rather than a hand-tuned constant.
Uses counterfactual causal learning to let an agent's harness adapt orchestration per task and environment, cutting long-horizon compute while holding success rate. One of the first papers to treat harness configuration as a learnable causal policy rather than a hand-tuned constant.
Orchestration and control flow are made explicit and inspectable. Uses counterfactual causal learning to let an agent's harness adapt orchestration per task and environment, cutting long-horizon compute while holding success rate. One of the first papers to treat harness configuration as a learnable causal policy rathe...
Use CHILL-Harness: Counterfactual Harness Learning for Efficient Reasoning in Long-Horizon Agents to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.25825; inspect its method and evaluation before treating results as production evidence.
medium
README.md
878
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L878
2026-07-30
Design
design
Specify a loop contract and operating pattern.
delegation
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.25825
[2607.25825] CHILL-Harness: Counterfactual Harness Learning for Efficient Reasoning in Long-Horizon Agents
Agent harnesses have become the operational infrastructure of modern large language model agents, coordinating context, tools, verification, and execution control to translate latent model capability into reliable long-horizon behavior. However, reliable long-horizon behavior requires harness control to adapt to task d...
Jiarun Fu; Lizhong Ding; Sida Chen; Honglei Xin; Chunhui Zhang; Pengqi Li; Qiuning Wei; Ye Yuan; Guoren Wang
2026-07-28
2026
arXiv
arXiv
cs.MA
arxiv-api
2607.25825
2026-08-13T14:24:04
ale-0257
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
A Control System, a Dataset, and a Recipe for Making Frozen LLM Agents Learn a Domain
https://arxiv.org/abs/2607.25415
external
arxiv.org
RL-optimizes the harness around a frozen model by treating it as a fixed action space rather than allowing unconstrained self-modification, evaluated across tool-use workflows, code generation, and retrieval QA with released code and data. The bounded-action-space framing is the safety-relevant counterpoint to open-end...
RL-optimizes the harness around a frozen model by treating it as a fixed action space rather than allowing unconstrained self-modification, evaluated across tool-use workflows, code generation, and retrieval QA with released code and data. The bounded-action-space framing is the safety-relevant counterpoint to open-end...
RL-optimizes the harness around a frozen model by treating it as a fixed action space rather than allowing unconstrained self-modification, evaluated across tool-use workflows, code generation, and retrieval QA with released code and data. The bounded-action-space framing is the safety-relevant counterpoint to open-end...
Packages the evidence as queryable CSV and JSONL rather than only a rendered page. RL-optimizes the harness around a frozen model by treating it as a fixed action space rather than allowing unconstrained self-modification, evaluated across tool-use workflows, code generation, and retrieval QA with released code and dat...
Use A Control System, a Dataset, and a Recipe for Making Frozen LLM Agents Learn a Domain to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.25415; inspect its method and evaluation before treating results as production evidence.
medium
README.md
879
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L879
2026-07-30
Design
design
Specify a loop contract and operating pattern.
workspace;context
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.25415
[2607.25415] A Control System, a Dataset, and a Recipe for Making Frozen LLM Agents Learn a Domain
Production LLM agents are increasingly assembled from a frozen model wrapped in a harness: a prompt template, a tool set, a memory/retrieval layer, a planning strategy, and a verification policy. Two 2026 systems, Meta-Harness (Lee et al., 2026) and HyperAgents (Meta AI, 2026), show that this harness can itself be opti...
Debjyoti Paul
2026-07-28
2026
arXiv
arXiv
8 pages, 1 figure, 3 tables. Code and dataset: https://github.com/dpaul0501/context-optimization-rl
cs.AI
arxiv-api
2607.25415
2026-08-13T14:24:04
ale-0258
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
Scores Are Not Decisions: Cost-Aware Stopping for Tool Acquisition in LLM Agents
https://arxiv.org/abs/2607.27083
external
arxiv.org
Frames how many tools an agent should acquire as an optimal-stopping problem, training on the gap between stopping now and optimal continuation and proving the objective aligns with the stopping target under heterogeneous costs. Directly addresses when to halt, one of the least-formalized parts of loop design.
Frames how many tools an agent should acquire as an optimal-stopping problem, training on the gap between stopping now and optimal continuation and proving the objective aligns with the stopping target under heterogeneous costs. Directly addresses when to halt, one of the least-formalized parts of loop design.
Frames how many tools an agent should acquire as an optimal-stopping problem, training on the gap between stopping now and optimal continuation and proving the objective aligns with the stopping target under heterogeneous costs. Directly addresses when to halt, one of the least-formalized parts of loop design.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Frames how many tools an agent should acquire as an optimal-stopping problem, training on the gap between stopping now and optimal continuation and proving the objective aligns with the stopping target under heterogeneous cost...
Use Scores Are Not Decisions: Cost-Aware Stopping for Tool Acquisition in LLM Agents to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.27083; inspect its method and evaluation before treating results as production evidence.
medium
README.md
880
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L880
2026-07-30
Design
design
Specify a loop contract and operating pattern.
objective;workspace;budget;exit
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.27083
[2607.27083] Scores Are Not Decisions: Cost-Aware Stopping for Tool Acquisition in LLM Agents
As LLM agents increasingly depend on diverse external services such as search engines, databases, and connectors, agent harnesses face a fundamental tool-selection challenge: acquiring too few tools leaves the task under-informed, while too many adds cost, context load, and privacy exposure. Routers and retrievers can ...
Yicheng Feng; Yan Zhang; Yan Cheng; Wei Qi
2026-07-29
2026
arXiv
arXiv
cs.LG
arxiv-api
2607.27083
2026-08-13T14:24:04
ale-0259
Agent Workflow Patterns
agent-workflow-patterns
Tool
🧰
Skill Recorder
https://github.com/microsoft/skill-recorder
external
github.com
Microsoft's tool for authoring loops from demonstration instead of from prose. It captures a real work session locally, clicks, app and window switches, pages visited, optional spoken narration, then uses GitHub Copilot CLI to reconstruct what you actually did as one intent plus an ordered step list you review and edit...
Microsoft's tool for authoring loops from demonstration instead of from prose. It captures a real work session locally, clicks, app and window switches, pages visited, optional spoken narration, then uses GitHub Copilot CLI to reconstruct what you actually did as one intent plus an ordered step list you review and edit...
Microsoft's tool for authoring loops from demonstration instead of from prose. It captures a real work session locally, clicks, app and window switches, pages visited, optional spoken narration, then uses GitHub Copilot CLI to reconstruct what you actually did as one intent plus an ordered step list you review and edit...
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Microsoft's tool for authoring loops from demonstration instead of from prose. It captures a real work session locally, clicks, app and window switches, pages visited, optional spoken narration, then uses GitHub Copilot CLI to...
Use Skill Recorder to turn a recurring-agent idea into an explicit loop contract.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
881
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L881
2026-08-02
Design
design
Specify a loop contract and operating pattern.
workspace
builder
workflow
enabling
source-implementation
A
ok
https://github.com/microsoft/skill-recorder
GitHub - microsoft/skill-recorder: Desktop app that records your on-screen work session and uses the GitHub Copilot CLI to reconstruct it as an intent + ordered steps, then builds a reusable Skill or Automation for Microsoft Scout, Microsoft Copilot Cowork, or Copilot Studio. · GitHub
Desktop app that records your on-screen work session and uses the GitHub Copilot CLI to reconstruct it as an intent + ordered steps, then builds a reusable Skill or Automation for Microsoft Scout, Microsoft Copilot Cowork, or Copilot Studio. - microsoft/skill-recorder
microsoft/skill-recorder
GitHub
html-meta
microsoft/skill-recorder
2026-08-13T14:24:04
ale-0260
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
SIGIL: Compiling Agent Skills into Typed Harnesses
https://arxiv.org/abs/2607.27309
external
arxiv.org
Identifies that prose skill files force an agent to re-derive control flow on every run, then introduces 'skill compilation', lowering prose skills into executable harnesses through a typed IR (AG-IR). Reports 86% vs 56% of mandated steps executed, 2.3x more completed procedures, and 42% fewer tokens, stable across mod...
Identifies that prose skill files force an agent to re-derive control flow on every run, then introduces 'skill compilation', lowering prose skills into executable harnesses through a typed IR (AG-IR). Reports 86% vs 56% of mandated steps executed, 2.3x more completed procedures, and 42% fewer tokens, stable across mod...
Identifies that prose skill files force an agent to re-derive control flow on every run, then introduces 'skill compilation', lowering prose skills into executable harnesses through a typed IR (AG-IR). Reports 86% vs 56% of mandated steps executed, 2.3x more completed procedures, and 42% fewer tokens, stable across mod...
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Identifies that prose skill files force an agent to re-derive control flow on every run, then introduces 'skill compilation', lowering prose skills into executable harnesses through a typed IR (AG-IR). Reports 86% vs 56% of ma...
Use SIGIL: Compiling Agent Skills into Typed Harnesses to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.27309; inspect its method and evaluation before treating results as production evidence.
medium
README.md
882
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L882
2026-08-02
Design
design
Specify a loop contract and operating pattern.
budget
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.27309
[2607.27309] SIGIL: Compiling Agent Skills into Typed Harnesses
AI-Integrated agents increasingly acquire capability from skills: prose procedure files loaded into a model's context and run by a tool-calling loop. A skill is described to the runtime but never encoded in it, so the model re-derives its control flow on every run and may skip mandated verification. Across 30 skills an...
Jayanaka Dantanarayana; Savini Kashmira; Lingjia Tang; Jason Mars
2026-07-29
2026
arXiv
arXiv
cs.SE
arxiv-api
2607.27309
2026-08-13T14:24:04
ale-0261
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
RecHarness: A Bandit-Routed Agentic Harness for Self-Evolving Recommender Systems
https://arxiv.org/abs/2607.29241
external
arxiv.org
Identifies a concrete instability in self-improving loops: when the LLM both picks the direction of a modification and generates the concrete hypothesis, search becomes unstable under limited experiment budgets.
Identifies a concrete instability in self-improving loops: when the LLM both picks the direction of a modification and generates the concrete hypothesis, search becomes unstable under limited experiment budgets.
Identifies a concrete instability in self-improving loops: when the LLM both picks the direction of a modification and generates the concrete hypothesis, search becomes unstable under limited experiment budgets.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Identifies a concrete instability in self-improving loops: when the LLM both picks the direction of a modification and generates the concrete hypothesis, search becomes unstable under limited experiment budgets.
Use RecHarness: A Bandit-Routed Agentic Harness for Self-Evolving Recommender Systems to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.29241; inspect its method and evaluation before treating results as production evidence.
medium
README.md
883
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L883
2026-08-05
Design
design
Specify a loop contract and operating pattern.
budget
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.29241
[2607.29241] RecHarness: A Bandit-Routed Agentic Harness for Self-Evolving Recommender Systems
Optimizing modern recommender models still depends heavily on engineers manually iterating over architectural, objective, and training-strategy changes. While LLM-based agents can automate this trial-and-error process, allowing the LLM to both select modification directions and generate concrete hypotheses often leads ...
Haoran Ling; Yuecheng Li; Zeyu Song; Jing Yao; Shuwen Kang; Chi Lu; Wenjin Wu; Peng Jiang
2026-07-31
2026
arXiv
arXiv
9 pages, 2 figures
cs.IR
arxiv-api
2607.29241
2026-08-13T14:24:04
ale-0262
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
DREvo: Distilling Recalibrated Historical Experience for Harness Self-Evolution
https://arxiv.org/abs/2607.26722
external
arxiv.org
Attacks the instability of harness self-evolution by anchoring evidence at function level, recalibrating which historical insights still apply given current state, and distilling role-conditioned search intent. Answers the practical question of when a past lesson should still steer the loop rather than mislead it.
Attacks the instability of harness self-evolution by anchoring evidence at function level, recalibrating which historical insights still apply given current state, and distilling role-conditioned search intent. Answers the practical question of when a past lesson should still steer the loop rather than mislead it.
Attacks the instability of harness self-evolution by anchoring evidence at function level, recalibrating which historical insights still apply given current state, and distilling role-conditioned search intent. Answers the practical question of when a past lesson should still steer the loop rather than mislead it.
State persistence is explicit enough for repeated runs and handoff. Attacks the instability of harness self-evolution by anchoring evidence at function level, recalibrating which historical insights still apply given current state, and distilling role-conditioned search intent. Answers the practical question of when a ...
Use DREvo: Distilling Recalibrated Historical Experience for Harness Self-Evolution to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.26722; inspect its method and evaluation before treating results as production evidence.
medium
README.md
884
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L884
2026-08-05
Design
design
Specify a loop contract and operating pattern.
state
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.26722
[2607.26722] DREvo: Distilling Recalibrated Historical Experience for Harness Self-Evolution
Harness plays a critical role in large language model agent performance, and building a high-performing harness requires substantial expert effort. Therefore, recent research has increasingly explored harness self-evolution, which iteratively proposes, evaluates, and improves harnesses using historical trial experience...
Hanghui Guo; Weijie Shi; Zhangze Chen; Shengxiang Xu; Yishu Wang; Yimei Zhang; Wangze Ni; Jia Zhu; Shimin Di
2026-07-29
2026
arXiv
arXiv
9 pages
cs.MA
arxiv-api
2607.26722
2026-08-13T14:24:04
ale-0263
Agent Workflow Patterns
agent-workflow-patterns
Paper
📄
SkillMentor: LLM Agent Self-Evolution via Learning Blind-Spot Diagnosis
https://arxiv.org/abs/2607.27360
external
arxiv.org
Treats blind-spot discovery as a learnable capability rather than assuming failures are already surfaced: an RL-trained Mentor policy generates diagnostic probe tasks, clusters recurring failure patterns, and converts them into reusable corrective skills without touching the executor or needing human labels.
Treats blind-spot discovery as a learnable capability rather than assuming failures are already surfaced: an RL-trained Mentor policy generates diagnostic probe tasks, clusters recurring failure patterns, and converts them into reusable corrective skills without touching the executor or needing human labels.
Treats blind-spot discovery as a learnable capability rather than assuming failures are already surfaced: an RL-trained Mentor policy generates diagnostic probe tasks, clusters recurring failure patterns, and converts them into reusable corrective skills without touching the executor or needing human labels.
Distills reusable agent-control patterns that are not tied to a single vendor implementation. Treats blind-spot discovery as a learnable capability rather than assuming failures are already surfaced: an RL-trained Mentor policy generates diagnostic probe tasks, clusters recurring failure patterns, and converts them int...
Use SkillMentor: LLM Agent Self-Evolution via Learning Blind-Spot Diagnosis to turn a recurring-agent idea into an explicit loop contract.
Research source arXiv:2607.27360; inspect its method and evaluation before treating results as production evidence.
medium
README.md
885
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L885
2026-08-05
Design
design
Specify a loop contract and operating pattern.
intake;escalation
researcher;evaluator
workflow
enabling
research-preprint
A
ok
https://arxiv.org/abs/2607.27360
[2607.27360] SkillMentor: LLM Agent Self-Evolution via Learning Blind-Spot Diagnosis
Agent self-evolution has primarily focused on learning how to act, while overlooking an equally important capability: learning to discover what an agent does not know. Existing approaches typically assume that failure discovery is given, focusing on how to repair failures once they are identified. We ask whether blind-...
Xiaoyi Bao; Yuanzhen Xie; Yunzhi Tan; Jinghang Gu; Zhongqing Wang; Chu-Ren Huang; Bo Hu; Zang Li
2026-07-29
2026
arXiv
arXiv
cs.AI
arxiv-api
2607.27360
2026-08-13T14:24:04
ale-0264
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
SWE-agent
https://github.com/SWE-agent/SWE-agent
external
github.com
Agent-computer interface and autonomous software engineering agent for repository tasks.
Agent-computer interface and autonomous software engineering agent for repository tasks.
Agent-computer interface and autonomous software engineering agent for repository tasks.
Uses real automated software-engineering systems as evidence for practical loop architectures. Agent-computer interface and autonomous software engineering agent for repository tasks.
Use SWE-agent to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
894
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L894
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
source-implementation
A
ok
https://github.com/SWE-agent/SWE-agent
GitHub - SWE-agent/SWE-agent: SWE-agent takes a GitHub issue and tries to automatically fix it, using your LM of choice. It can also be employed for offensive cybersecurity or competitive coding challenges. [NeurIPS 2024] · GitHub
SWE-agent takes a GitHub issue and tries to automatically fix it, using your LM of choice. It can also be employed for offensive cybersecurity or competitive coding challenges. [NeurIPS 2024] - GitHub - SWE-agent/SWE-agent: SWE-agent takes a GitHub issue and tries to automatically fix it, using your LM of choice. It ca...
SWE-agent/SWE-agent
GitHub
html-meta
SWE-agent/SWE-agent
2026-08-13T14:24:04
ale-0265
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering
https://arxiv.org/abs/2405.15793
external
arxiv.org
Paper behind SWE-agent and its interface design.
Paper behind SWE-agent and its interface design.
Paper behind SWE-agent and its interface design.
Uses real automated software-engineering systems as evidence for practical loop architectures. Paper behind SWE-agent and its interface design.
Use SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering to choose an implementation surface for repeatable agent work.
Research source arXiv:2405.15793; inspect its method and evaluation before treating results as production evidence.
medium
README.md
895
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L895
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
researcher;evaluator
agent
direct
research-paper
A
ok
https://proceedings.neurips.cc/paper_files/paper/2024/hash/5a7c947568c1b1328ccc5230172e1e7c-Abstract-Conference.html
[2405.15793] SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering
Language model (LM) agents are increasingly being used to automate complicated tasks in digital environments. Just as humans benefit from powerful software applications, such as integrated development environments, for complex tasks like software engineering, we posit that LM agents represent a new category of end user...
John Yang; Carlos E. Jimenez; Alexander Wettig; Kilian Lieret; Shunyu Yao; Karthik Narasimhan; Ofir Press
2024
2024
Advances in Neural Information Processing Systems 37 (NeurIPS)
Neural Information Processing Systems Foundation
10.52202/079017-1601
Published in Advances in Neural Information Processing Systems 37 (NeurIPS); the linked arXiv record remains available for open access.
cs.SE
NeurIPS proceedings and DOI records
2405.15793
2026-08-13T14:24:04
ale-0266
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
mini-SWE-agent
https://mini-swe-agent.com/latest/
external
mini-swe-agent.com
Minimal coding agent that is useful for understanding the core loop without a large framework.
Minimal coding agent that is useful for understanding the core loop without a large framework.
Minimal coding agent that is useful for understanding the core loop without a large framework.
Uses real automated software-engineering systems as evidence for practical loop architectures. Minimal coding agent that is useful for understanding the core loop without a large framework.
Use mini-SWE-agent to choose an implementation surface for repeatable agent work.
Working implementation, framework, runtime, or repository; signal comes from usable code and ecosystem adoption.
high
README.md
896
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L896
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
implementation
A
ok
https://mini-swe-agent.com/latest/
Overview - mini-SWE-agent documentation
mini-swe-agent.com
domain-fallback
2026-08-13T14:24:04
ale-0267
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
OpenHands
https://github.com/All-Hands-AI/OpenHands
external
github.com
Open platform for AI software developers as generalist agents.
Open platform for AI software developers as generalist agents.
Open platform for AI software developers as generalist agents.
Uses real automated software-engineering systems as evidence for practical loop architectures. Open platform for AI software developers as generalist agents.
Use OpenHands to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
897
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L897
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
source-implementation
A
ok
https://github.com/OpenHands/OpenHands
GitHub - OpenHands/OpenHands: 🙌 OpenHands: AI-Driven Development · GitHub
🙌 OpenHands: AI-Driven Development. Contribute to OpenHands/OpenHands development by creating an account on GitHub.
All-Hands-AI/OpenHands
GitHub
html-meta
All-Hands-AI/OpenHands
2026-08-13T14:24:04
ale-0268
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
OpenHands: An Open Platform for AI Software Developers as Generalist Agents
https://arxiv.org/abs/2407.16741
external
arxiv.org
Paper describing OpenHands, CodeActAgent, benchmarks, and generalist agent evaluation.
Paper describing OpenHands, CodeActAgent, benchmarks, and generalist agent evaluation.
Paper describing OpenHands, CodeActAgent, benchmarks, and generalist agent evaluation.
Evaluation data is used as the feedback signal for improving loop behavior. Paper describing OpenHands, CodeActAgent, benchmarks, and generalist agent evaluation.
Use OpenHands: An Open Platform for AI Software Developers as Generalist Agents to choose an implementation surface for repeatable agent work.
Research source arXiv:2407.16741; inspect its method and evaluation before treating results as production evidence.
medium
README.md
898
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L898
Build
build
Choose runtimes, tools, and delegation surfaces.
verification
researcher;evaluator
agent
direct
research-paper
A
ok
https://proceedings.iclr.cc/paper_files/paper/2025/hash/a4b6ad6b48850c0c331d1259fc66a69c-Abstract-Conference.html
[2407.16741] OpenHands: An Open Platform for AI Software Developers as Generalist Agents
Software is one of the most powerful tools that we humans have at our disposal; it allows a skilled programmer to interact with the world in complex and profound ways. At the same time, thanks to improvements in large language models (LLMs), there has also been a rapid development in AI agents that interact with and af...
Xingyao Wang; Boxuan Li; Yufan Song; Frank F. Xu; Xiangru Tang; Mingchen Zhuge; Jiayi Pan; Yueqi Song; Bowen Li; Jaskirat Singh; Hoang H. Tran; Fuqiang Li; Ren Ma; Mingzhang Zheng; Bill Qian; Yanjun Shao; Niklas Muennighoff; Yizhe Zhang; Binyuan Hui; Junyang Lin; Robert Brennan; Hao Peng; Heng Ji; Graham Neubig
2025
2025
International Conference on Learning Representations (ICLR)
International Conference on Learning Representations
Published in International Conference on Learning Representations (ICLR); the linked arXiv record remains available for open access.
cs.SE
ICLR proceedings record
2407.16741
2026-08-13T14:24:04
ale-0269
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
Agentless
https://github.com/OpenAutoCoder/Agentless
external
github.com
Workflow-based approach for software issue resolution using localization, repair, and patch validation.
Workflow-based approach for software issue resolution using localization, repair, and patch validation.
Workflow-based approach for software issue resolution using localization, repair, and patch validation.
Uses real automated software-engineering systems as evidence for practical loop architectures. Workflow-based approach for software issue resolution using localization, repair, and patch validation.
Use Agentless to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
899
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L899
Build
build
Choose runtimes, tools, and delegation surfaces.
intake
builder
agent
direct
source-implementation
A
ok
https://github.com/OpenAutoCoder/Agentless
GitHub - OpenAutoCoder/Agentless: Agentless🐱: an agentless approach to automatically solve software development problems · GitHub
Agentless🐱: an agentless approach to automatically solve software development problems - OpenAutoCoder/Agentless
OpenAutoCoder/Agentless
GitHub
html-meta
OpenAutoCoder/Agentless
2026-08-13T14:24:04
ale-0270
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
Agentless: Demystifying LLM-based Software Engineering Agents
https://arxiv.org/abs/2407.01489
external
arxiv.org
Useful contrast case: strong results through structured workflow rather than a fully open-ended agent.
Useful contrast case: strong results through structured workflow rather than a fully open-ended agent.
Useful contrast case: strong results through structured workflow rather than a fully open-ended agent.
Uses real automated software-engineering systems as evidence for practical loop architectures. Useful contrast case: strong results through structured workflow rather than a fully open-ended agent.
Use Agentless: Demystifying LLM-based Software Engineering Agents to choose an implementation surface for repeatable agent work.
Research source arXiv:2407.01489; inspect its method and evaluation before treating results as production evidence.
medium
README.md
900
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L900
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2407.01489
[2407.01489] Agentless: Demystifying LLM-based Software Engineering Agents
Recent advancements in large language models (LLMs) have significantly advanced the automation of software development tasks, including code synthesis, program repair, and test generation. More recently, researchers and industry practitioners have developed various autonomous LLM agents to perform end-to-end software d...
Chunqiu Steven Xia; Yinlin Deng; Soren Dunn; Lingming Zhang
2024-07-01
2024
arXiv
arXiv
cs.SE
arxiv-api
2407.01489
2026-08-13T14:24:04
ale-0271
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
AutoCodeRover
https://github.com/AutoCodeRoverSG/auto-code-rover
external
github.com
Autonomous program improvement system for issue localization, patch generation, and validation.
Autonomous program improvement system for issue localization, patch generation, and validation.
Autonomous program improvement system for issue localization, patch generation, and validation.
Uses real automated software-engineering systems as evidence for practical loop architectures. Autonomous program improvement system for issue localization, patch generation, and validation.
Use AutoCodeRover to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
901
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L901
Build
build
Choose runtimes, tools, and delegation surfaces.
intake
builder
agent
direct
source-implementation
A
ok
https://github.com/AutoCodeRoverSG/auto-code-rover
GitHub - AutoCodeRoverSG/auto-code-rover: A project structure aware autonomous software engineer aiming for autonomous program improvement. Resolved 37.3% tasks (pass@1) in SWE-bench lite and 46.2% tasks (pass@1) in SWE-bench verified with each task costs less than $0.7. · GitHub
A project structure aware autonomous software engineer aiming for autonomous program improvement. Resolved 37.3% tasks (pass@1) in SWE-bench lite and 46.2% tasks (pass@1) in SWE-bench verified with each task costs less than $0.7. - AutoCodeRoverSG/auto-code-rover
AutoCodeRoverSG/auto-code-rover
GitHub
html-meta
AutoCodeRoverSG/auto-code-rover
2026-08-13T14:24:04
ale-0272
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
AutoCodeRover: Autonomous Program Improvement
https://arxiv.org/abs/2404.05427
external
arxiv.org
Paper on autonomous code repair loops over real repositories.
Paper on autonomous code repair loops over real repositories.
Paper on autonomous code repair loops over real repositories.
Uses real automated software-engineering systems as evidence for practical loop architectures. Paper on autonomous code repair loops over real repositories.
Use AutoCodeRover: Autonomous Program Improvement to choose an implementation surface for repeatable agent work.
Research source arXiv:2404.05427; inspect its method and evaluation before treating results as production evidence.
medium
README.md
902
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L902
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
researcher;evaluator
agent
direct
research-paper
A
ok
https://doi.org/10.1145/3650212.3680384
[2404.05427] AutoCodeRover: Autonomous Program Improvement
Researchers have made significant progress in automating the software development process in the past decades. Recent progress in Large Language Models (LLMs) has significantly impacted the development process, where developers can use LLM-based programming assistants to achieve automated coding. Nevertheless, software...
Yuntong Zhang; Haifeng Ruan; Zhiyu Fan; Abhik Roychoudhury
2024-09-11
2024
Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA)
Association for Computing Machinery
10.1145/3650212.3680384
Published in Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA); the linked arXiv record remains available for open access.
cs.SE
ACM DOI record
2404.05427
2026-08-13T14:24:04
ale-0273
Coding-Agent Loop Systems
coding-agent-loop-systems
List
🧭
SWE-bench reading list
https://github.com/SWE-bench/reading-list
external
github.com
Maintained map of software engineering agent systems and related papers.
Maintained map of software engineering agent systems and related papers.
Maintained map of software engineering agent systems and related papers.
Uses real automated software-engineering systems as evidence for practical loop architectures. Maintained map of software engineering agent systems and related papers.
Use SWE-bench reading list to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
903
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L903
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
curated-index
C
ok
https://github.com/SWE-bench/reading-list
GitHub - SWE-bench/reading-list: Academic papers and works related to SWE-bench and SWE-agents · GitHub
Academic papers and works related to SWE-bench and SWE-agents - SWE-bench/reading-list
SWE-bench/reading-list
GitHub
html-meta
SWE-bench/reading-list
2026-08-13T14:24:04
ale-0274
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
TraceCoder: A Trace-Driven Multi-Agent Framework for Automated Debugging of LLM-Generated Code
https://arxiv.org/abs/2602.06875
external
arxiv.org
ICSE'26 observe-analyze-repair loop with instrumentation, analysis, and repair agents, a history-learning mechanism, and a rollback to the last good state; iteration alone drives most of the gain.
ICSE'26 observe-analyze-repair loop with instrumentation, analysis, and repair agents, a history-learning mechanism, and a rollback to the last good state; iteration alone drives most of the gain.
ICSE'26 observe-analyze-repair loop with instrumentation, analysis, and repair agents, a history-learning mechanism, and a rollback to the last good state; iteration alone drives most of the gain.
The work separates roles across agents, verifiers, or orchestration layers. ICSE'26 observe-analyze-repair loop with instrumentation, analysis, and repair agents, a history-learning mechanism, and a rollback to the last good state; iteration alone drives most of the gain.
Use TraceCoder: A Trace-Driven Multi-Agent Framework for Automated Debugging of LLM-Generated Code to choose an implementation surface for repeatable agent work.
Research source arXiv:2602.06875; inspect its method and evaluation before treating results as production evidence.
medium
README.md
904
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L904
Build
build
Choose runtimes, tools, and delegation surfaces.
delegation;state
researcher;evaluator
agent
direct
research-paper
A
ok
https://conf.researchr.org/details/icse-2026/icse-2026-research-track/145/TraceCoder-A-Trace-Driven-Multi-Agent-Framework-for-Automated-Debugging-of-LLM-Gener
[2602.06875] TraceCoder: A Trace-Driven Multi-Agent Framework for Automated Debugging of LLM-Generated Code
Large Language Models (LLMs) often generate code with subtle but critical bugs, especially for complex tasks. Existing automated repair methods typically rely on superficial pass/fail signals, offering limited visibility into program behavior and hindering precise error localization. In addition, without a way to learn...
Jiangping Huang; Wenguang Ye; Weisong Sun; Jian Zhang; Mingyue Zhang; Yang Liu
2026-04-12
2026
Proceedings of the 48th IEEE/ACM International Conference on Software Engineering (ICSE)
Association for Computing Machinery
10.1145/3744916.3773187
Published in Proceedings of the 48th IEEE/ACM International Conference on Software Engineering (ICSE); the linked arXiv record remains available for open access.
cs.SE
ICSE program and camera-ready records
2602.06875
2026-08-13T14:24:04
ale-0275
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
The Kitchen Loop: User-Spec-Driven Development for a Self-Evolving Codebase
https://arxiv.org/abs/2603.25697
external
arxiv.org
Production loop where an agent exercises a spec surface as a synthetic power user behind ground-truth tests and quality gates, reporting 285+ self-correcting iterations and 1,000+ merged PRs with zero detected regressions.
Production loop where an agent exercises a spec surface as a synthetic power user behind ground-truth tests and quality gates, reporting 285+ self-correcting iterations and 1,000+ merged PRs with zero detected regressions.
Production loop where an agent exercises a spec surface as a synthetic power user behind ground-truth tests and quality gates, reporting 285+ self-correcting iterations and 1,000+ merged PRs with zero detected regressions.
Uses real automated software-engineering systems as evidence for practical loop architectures. Production loop where an agent exercises a spec surface as a synthetic power user behind ground-truth tests and quality gates, reporting 285+ self-correcting iterations and 1,000+ merged PRs with zero detected regressions.
Use The Kitchen Loop: User-Spec-Driven Development for a Self-Evolving Codebase to choose an implementation surface for repeatable agent work.
Research source arXiv:2603.25697; inspect its method and evaluation before treating results as production evidence.
medium
README.md
905
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L905
Build
build
Choose runtimes, tools, and delegation surfaces.
verification
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2603.25697
[2603.25697] The Kitchen Loop: User-Spec-Driven Development for a Self-Evolving Codebase
Code production is now a commodity; the bottleneck is knowing what to build and proving it works. We present the Kitchen Loop, a framework for autonomous, self-evolving software built on a unified trust model: (1) a specification surface enumerating what the product claims to support; (2) 'As a User x 1000', where an L...
Yannick Roy
2026-03-26
2026
arXiv
arXiv
cs.SE
arxiv-api
2603.25697
2026-08-13T14:24:04
ale-0276
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
Inside the Scaffold: A Source-Code Taxonomy of Coding Agent Architectures
https://arxiv.org/abs/2604.03515
external
arxiv.org
Dissects 13 open-source coding-agent scaffolds and identifies five composable loop primitives (ReAct, generate-test-repair, plan-execute, retry, tree search) that real agents layer, mapping how control loop, tools, and state combine.
Dissects 13 open-source coding-agent scaffolds and identifies five composable loop primitives (ReAct, generate-test-repair, plan-execute, retry, tree search) that real agents layer, mapping how control loop, tools, and state combine.
Dissects 13 open-source coding-agent scaffolds and identifies five composable loop primitives (ReAct, generate-test-repair, plan-execute, retry, tree search) that real agents layer, mapping how control loop, tools, and state combine.
State persistence is explicit enough for repeated runs and handoff. Dissects 13 open-source coding-agent scaffolds and identifies five composable loop primitives (ReAct, generate-test-repair, plan-execute, retry, tree search) that real agents layer, mapping how control loop, tools, and state combine.
Use Inside the Scaffold: A Source-Code Taxonomy of Coding Agent Architectures to choose an implementation surface for repeatable agent work.
Research source arXiv:2604.03515; inspect its method and evaluation before treating results as production evidence.
medium
README.md
906
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L906
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;verification;state;budget
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2604.03515
[2604.03515] Inside the Scaffold: A Source-Code Taxonomy of Coding Agent Architectures
LLM-based coding agents can localize bugs, generate patches, and run tests with diminishing human oversight, yet the scaffolding code that surrounds the language model (the control loop, tool definitions, state management, and context strategy) remains poorly understood. Existing surveys classify agents by abstract cap...
Benjamin Rombaut
2026-04-03
2026
arXiv
arXiv
cs.SE
arxiv-api
2604.03515
2026-08-13T14:24:04
ale-0277
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
A Self-Improving Coding Agent
https://arxiv.org/abs/2504.15228
external
arxiv.org
An agent that edits its own code and tools and re-runs against a benchmark, lifting itself from 17% to 53% on a SWE-bench Verified subset, a concrete self-modifying improvement loop.
An agent that edits its own code and tools and re-runs against a benchmark, lifting itself from 17% to 53% on a SWE-bench Verified subset, a concrete self-modifying improvement loop.
An agent that edits its own code and tools and re-runs against a benchmark, lifting itself from 17% to 53% on a SWE-bench Verified subset, a concrete self-modifying improvement loop.
Verification is promoted from a final check to a loop-control signal. An agent that edits its own code and tools and re-runs against a benchmark, lifting itself from 17% to 53% on a SWE-bench Verified subset, a concrete self-modifying improvement loop.
Use A Self-Improving Coding Agent to choose an implementation surface for repeatable agent work.
Research source arXiv:2504.15228; inspect its method and evaluation before treating results as production evidence.
medium
README.md
907
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L907
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;verification
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2504.15228
[2504.15228] A Self-Improving Coding Agent
Recent advancements in Large Language Models (LLMs) have spurred interest in deploying LLM agents to undertake tasks in the world. LLMs are often deployed in agent systems: code that orchestrates LLM calls and provides them with tools. We demonstrate that an agent system, equipped with basic coding tools, can autonomou...
Maxime Robeyns; Martin Szummer; Laurence Aitchison
2025-04-21
2025
arXiv
arXiv
Submitted as a preprint to NeurIPS 2025
cs.AI
arxiv-api
2504.15228
2026-08-13T14:24:04
ale-0278
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
Don't Blame the Large Language Model: How Scaffolding Evolution Shapes Coding Agent Quality
https://arxiv.org/abs/2607.03691
external
arxiv.org
Longitudinal study of 35 Qwen Code CLI releases with the model held constant, tracing coding-agent shifts to specific scaffolding changes in system prompts, tools, context management, and reasoning loops, and separating scaffolding regressions from model regressions.
Longitudinal study of 35 Qwen Code CLI releases with the model held constant, tracing coding-agent shifts to specific scaffolding changes in system prompts, tools, context management, and reasoning loops, and separating scaffolding regressions from model regressions.
Longitudinal study of 35 Qwen Code CLI releases with the model held constant, tracing coding-agent shifts to specific scaffolding changes in system prompts, tools, context management, and reasoning loops, and separating scaffolding regressions from model regressions.
Context is managed as durable loop state rather than a single prompt payload. Longitudinal study of 35 Qwen Code CLI releases with the model held constant, tracing coding-agent shifts to specific scaffolding changes in system prompts, tools, context management, and reasoning loops, and separating scaffolding regression...
Use Don't Blame the Large Language Model: How Scaffolding Evolution Shapes Coding Agent Quality to choose an implementation surface for repeatable agent work.
Research source arXiv:2607.03691; inspect its method and evaluation before treating results as production evidence.
medium
README.md
908
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L908
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;context
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2607.03691
[2607.03691] Don't Blame the Large Language Model: How Agent Harness Evolution Shapes Coding Agent Quality
Coding agents, autonomous systems that use large language models (LLMs) to resolve software engineering tasks, rely on agent harness: a middleware layer in between a developer and a large language model that orchestrates system prompts, tool execution, context management, and iterative reasoning loops. While these agen...
Oussama Ben Sghaier; Hao Li; Bram Adams; Ahmed E. Hassan
2026-07-04
2026
arXiv
arXiv
cs.SE
arxiv-api
2607.03691
2026-08-13T14:24:04
ale-0279
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
ToFu: A White-Box, Token-Efficient Agent Harness for Researchers
https://arxiv.org/abs/2607.11423
external
arxiv.org
MIT-licensed white-box agent harness built on the thesis that agent behavior is set by the orchestration code around the model as much as the model itself, letting researchers inspect, modify, and evaluate its orchestration logic with reported token-efficiency gains over existing harnesses.
MIT-licensed white-box agent harness built on the thesis that agent behavior is set by the orchestration code around the model as much as the model itself, letting researchers inspect, modify, and evaluate its orchestration logic with reported token-efficiency gains over existing harnesses.
MIT-licensed white-box agent harness built on the thesis that agent behavior is set by the orchestration code around the model as much as the model itself, letting researchers inspect, modify, and evaluate its orchestration logic with reported token-efficiency gains over existing harnesses.
Orchestration and control flow are made explicit and inspectable. MIT-licensed white-box agent harness built on the thesis that agent behavior is set by the orchestration code around the model as much as the model itself, letting researchers inspect, modify, and evaluate its orchestration logic with reported token-effi...
Use ToFu: A White-Box, Token-Efficient Agent Harness for Researchers to choose an implementation surface for repeatable agent work.
Research source arXiv:2607.11423; inspect its method and evaluation before treating results as production evidence.
medium
README.md
909
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L909
2026-07-15
Build
build
Choose runtimes, tools, and delegation surfaces.
delegation;budget
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2607.11423
[2607.11423] ToFu: A White-Box, Token-Efficient Agent Harness for Researchers
Agentic coding tools present new opportunities to transform research workflows. The performance of agent systems built depends on both large language models (LLMs) and the harness around LLMs, which is the orchestration code that determines an agent's behavior. We present ToFu, an agentic harness for researchers that r...
Junhao Ruan; Yuan Ge; Bei Li; Yongjing Yin; Yuchun Fan; Xin Chen; Jingang Wang; Chenglong Wang; Jingbo Zhu; Tong Xiao
2026-07-13
2026
arXiv
arXiv
cs.CL
arxiv-api
2607.11423
2026-08-13T14:24:04
ale-0280
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
When Does Restricting a Coding Agent to execute_code Help?
https://arxiv.org/abs/2607.10569
external
arxiv.org
Regime-by-design ablation measuring when constraining a coding agent to a single execute_code action helps or hurts, separating task regime from agent design so harness choices can be made from evidence rather than intuition.
Regime-by-design ablation measuring when constraining a coding agent to a single execute_code action helps or hurts, separating task regime from agent design so harness choices can be made from evidence rather than intuition.
Regime-by-design ablation measuring when constraining a coding agent to a single execute_code action helps or hurts, separating task regime from agent design so harness choices can be made from evidence rather than intuition.
Uses real automated software-engineering systems as evidence for practical loop architectures. Regime-by-design ablation measuring when constraining a coding agent to a single execute_code action helps or hurts, separating task regime from agent design so harness choices can be made from evidence rather than intuition.
Use When Does Restricting a Coding Agent to execute_code Help? to choose an implementation surface for repeatable agent work.
Research source arXiv:2607.10569; inspect its method and evaluation before treating results as production evidence.
medium
README.md
910
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L910
2026-07-15
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
researcher;evaluator
agent
direct
research-paper
A
ok
https://arxiv.org/abs/2607.10569
[2607.10569] When Does Restricting a Coding Agent to execute_code Help? A Regime $\times$ Agent-Design Ablation
Modern coding agents expose multiple tool surfaces -- IDE primitives, bash, and Model Context Protocol (MCP) code-execution -- and the field has shipped three contradictory claims about which one matters. We run the missing crossed comparison: an integrity-clean three-arm ablation (baseline / bash_only / code_only) on ...
Hong Yang; Qi Yu; Travis Desell
2026
2026
KDD Workshop on Agentic Software Engineering (SE 3.0)
ACM SIGKDD
Accepted at KDD Workshop on Agentic Software Engineering (SE 3.0); the linked arXiv record is the available paper version.
cs.SE
Current arXiv acceptance note and official non-archival workshop page
2607.10569
2026-08-13T14:24:04
ale-0281
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
Agentic Synthesis against Counterexample-Supplemented Sketches
https://arxiv.org/abs/2607.15854
external
arxiv.org
Turns human-approved failures into durable policy: a coding agent revises a code-shaped sketch, regenerates code and prompt surfaces, preserves provenance, and must pass a regression gate before seeing the next case.
Turns human-approved failures into durable policy: a coding agent revises a code-shaped sketch, regenerates code and prompt surfaces, preserves provenance, and must pass a regression gate before seeing the next case.
Turns human-approved failures into durable policy: a coding agent revises a code-shaped sketch, regenerates code and prompt surfaces, preserves provenance, and must pass a regression gate before seeing the next case.
Durable execution and replay are treated as first-class loop infrastructure. Turns human-approved failures into durable policy: a coding agent revises a code-shaped sketch, regenerates code and prompt surfaces, preserves provenance, and must pass a regression gate before seeing the next case.
Use Agentic Synthesis against Counterexample-Supplemented Sketches to choose an implementation surface for repeatable agent work.
Research source arXiv:2607.15854; inspect its method and evaluation before treating results as production evidence.
medium
README.md
911
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L911
2026-07-20
Build
build
Choose runtimes, tools, and delegation surfaces.
escalation
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2607.15854
[2607.15854] Agentic Synthesis against Counterexample-Supplemented Sketches
Coding agents can fix a failing example without preserving the domain rule that made it fail. We present agentic synthesis against counterexample-supplemented sketches, a repository-native method for systems whose policy is discovered during implementation. A human starts with a partial sketch, and a coding agent compi...
Muness Castle; Eric Rubeck
2026-07-17
2026
arXiv
arXiv
32 pages, 5 displayed figures (4 distinct screenshots). Includes the CatSynth artifact supplement. Code and captured experiment artifacts: https://github.com/open-horizon-labs/counterexample-supplemented-sketches Clarifies the two-check CESS method and Developer change authority; adds the protocol-correct CatSynth reru...
cs.SE
arxiv-api
2607.15854
2026-08-13T14:24:04
ale-0282
Coding-Agent Loop Systems
coding-agent-loop-systems
Paper
📄
Making Agent-Mediated Contributions Governable: A Project-Level Governance Manifest for Open-Source AI Collaboration
https://arxiv.org/abs/2607.15769
external
arxiv.org
Defines a repository-hosted governance contract connecting contributor evidence with maintainer verification and decision authority; after an audit of 50 repositories, a 15-reviewer study improves exact risk-label recovery from 15/37 to 37/38 with the manifest-supported materials.
Defines a repository-hosted governance contract connecting contributor evidence with maintainer verification and decision authority; after an audit of 50 repositories, a 15-reviewer study improves exact risk-label recovery from 15/37 to 37/38 with the manifest-supported materials.
Defines a repository-hosted governance contract connecting contributor evidence with maintainer verification and decision authority; after an audit of 50 repositories, a 15-reviewer study improves exact risk-label recovery from 15/37 to 37/38 with the manifest-supported materials.
Verification is promoted from a final check to a loop-control signal. Defines a repository-hosted governance contract connecting contributor evidence with maintainer verification and decision authority; after an audit of 50 repositories, a 15-reviewer study improves exact risk-label recovery from 15/37 to 37/38 with th...
Use Making Agent-Mediated Contributions Governable: A Project-Level Governance Manifest for Open-Source AI Collaboration to choose an implementation surface for repeatable agent work.
Research source arXiv:2607.15769; inspect its method and evaluation before treating results as production evidence.
medium
README.md
912
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L912
2026-07-20
Build
build
Choose runtimes, tools, and delegation surfaces.
verification
researcher;evaluator
agent
direct
research-preprint
A
ok
https://arxiv.org/abs/2607.15769
[2607.15769] Making Agent-Mediated Contributions Governable: A Project-Level Governance Manifest for Open-Source AI Collaboration
Generative AI and coding agents are intensifying a central governance tension in open-source software (OSS): they scale contribution generation faster than maintainers can assess risk, evidence, and accountability. Existing responses improve agent-readability and traceability, but project rules must also organize contr...
Jinjin Gao; Luyang Li; Shufen Guo; Ligang He; Xiaoning Sun
2026-07-17
2026
arXiv
arXiv
Preprint. Under journal review
cs.SE
arxiv-api
2607.15769
2026-08-13T14:24:04
ale-0283
Coding-Agent Loop Systems
coding-agent-loop-systems
Pattern
🔁
Ralph
https://ghuntley.com/ralph/
external
ghuntley.com
Geoffrey Huntley's original Ralph technique: run one agent in a bare loop with fresh context per iteration and the filesystem plus specs as memory.
Geoffrey Huntley's original Ralph technique: run one agent in a bare loop with fresh context per iteration and the filesystem plus specs as memory.
Geoffrey Huntley's original Ralph technique: run one agent in a bare loop with fresh context per iteration and the filesystem plus specs as memory.
Persistent memory is treated as an external runtime artifact. Geoffrey Huntley's original Ralph technique: run one agent in a bare loop with fresh context per iteration and the filesystem plus specs as memory.
Use Ralph to choose an implementation surface for repeatable agent work.
Operational pattern or playbook; signal comes from reusable loop structure and practical transferability.
medium
README.md
917
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L917
Build
build
Choose runtimes, tools, and delegation surfaces.
context
builder
agent
direct
operational-pattern
B
ok
https://ghuntley.com/ralph/
Ralph Wiggum as a "software engineer"
How Ralph Wiggum went from 'The Simpsons' to the biggest name in AI right now - Venture Beat 😎Here's a cool little field report from a Y Combinator hackathon event where they put Ralph Wiggum to the test. "We Put a Coding Agent in a While Loop and It Shipped
2025-07-14
2025
Geoffrey Huntley
html-meta
2026-08-13T14:24:04
ale-0284
Coding-Agent Loop Systems
coding-agent-loop-systems
Pattern
🔁
everything is a ralph loop
https://ghuntley.com/loop/
external
ghuntley.com
Follow-up essay arguing the loop, not the agent, is the durable engineering unit: one task per iteration, deterministic context, and verification inside the loop.
Follow-up essay arguing the loop, not the agent, is the durable engineering unit: one task per iteration, deterministic context, and verification inside the loop.
Follow-up essay arguing the loop, not the agent, is the durable engineering unit: one task per iteration, deterministic context, and verification inside the loop.
Durable execution and replay are treated as first-class loop infrastructure. Follow-up essay arguing the loop, not the agent, is the durable engineering unit: one task per iteration, deterministic context, and verification inside the loop.
Use everything is a ralph loop to choose an implementation surface for repeatable agent work.
Operational pattern or playbook; signal comes from reusable loop structure and practical transferability.
medium
README.md
918
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L918
Build
build
Choose runtimes, tools, and delegation surfaces.
context;verification
builder
agent
direct
operational-pattern
B
ok
https://ghuntley.com/loop/
everything is a ralph loop
I’ve been thinking about how I build software is so very very different how I used to do it three years ago. No, I’m not talking about acceleration through usage of AI but instead at a more fundamental level of approach, techniques and best practices. Standard software practices
2026-01-17
2026
Geoffrey Huntley
html-meta
2026-08-13T14:24:04
ale-0285
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
how-to-ralph-wiggum
https://github.com/ghuntley/how-to-ralph-wiggum
external
github.com
Reference repository documenting the Ralph Wiggum technique end to end, from the bare loop script to guardrails and conventions.
Reference repository documenting the Ralph Wiggum technique end to end, from the bare loop script to guardrails and conventions.
Reference repository documenting the Ralph Wiggum technique end to end, from the bare loop script to guardrails and conventions.
Uses real automated software-engineering systems as evidence for practical loop architectures. Reference repository documenting the Ralph Wiggum technique end to end, from the bare loop script to guardrails and conventions.
Use how-to-ralph-wiggum to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
919
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L919
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
source-implementation
A
ok
https://github.com/ghuntley/how-to-ralph-wiggum
GitHub - ghuntley/how-to-ralph-wiggum: The Ralph Wiggum Technique—the AI development methodology that reduces software costs to less than a fast food worker's wage. · GitHub
The Ralph Wiggum Technique—the AI development methodology that reduces software costs to less than a fast food worker's wage. - ghuntley/how-to-ralph-wiggum
ghuntley/how-to-ralph-wiggum
GitHub
html-meta
ghuntley/how-to-ralph-wiggum
2026-08-13T14:24:04
ale-0286
Coding-Agent Loop Systems
coding-agent-loop-systems
Blog
📝
A Brief History of Ralph
https://www.humanlayer.dev/blog/brief-history-of-ralph
external
www.humanlayer.dev
Traces how the bare-loop technique spread from a provocation to a production practice among early adopters.
Traces how the bare-loop technique spread from a provocation to a production practice among early adopters.
Traces how the bare-loop technique spread from a provocation to a production practice among early adopters.
Uses real automated software-engineering systems as evidence for practical loop architectures. Traces how the bare-loop technique spread from a provocation to a production practice among early adopters.
Use A Brief History of Ralph to choose an implementation surface for repeatable agent work.
Contextual source from www.humanlayer.dev; useful for practice signals or boundary conditions, not independent validation.
contextual
README.md
920
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L920
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
practitioner-analysis
B
ok
https://www.humanlayer.dev/blog/brief-history-of-ralph
A Brief History of Ralph | HumanLayer Blog
The Ralph Wiggum Technique went viral in the last week of 2025. Here's the story of ralph since the first time I met Geoff in June of 2025.
2026
2026
humanlayer.dev
html-meta
2026-08-13T14:24:04
ale-0287
Coding-Agent Loop Systems
coding-agent-loop-systems
Pattern
🔁
Ralph Copilot
https://github.com/giocaizzi/ralph-copilot/tree/e5b2813cc876c73a8c9d3398c0115da0d15f63cf
external
github.com
Language-agnostic Ralph loop implementation using fresh context, filesystem memory, `PRD.md`, and `PROGRESS.md`.
Language-agnostic Ralph loop implementation using fresh context, filesystem memory, `PRD.md`, and `PROGRESS.md`.
Language-agnostic Ralph loop implementation using fresh context, filesystem memory, `PRD.md`, and `PROGRESS.md`.
Persistent memory is treated as an external runtime artifact. Language-agnostic Ralph loop implementation using fresh context, filesystem memory, `PRD.md`, and `PROGRESS.md`.
Use Ralph Copilot to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
921
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L921
Build
build
Choose runtimes, tools, and delegation surfaces.
context
builder
agent
direct
operational-pattern
B
ok
https://github.com/giocaizzi/ralph-copilot/tree/e5b2813cc876c73a8c9d3398c0115da0d15f63cf
GitHub - giocaizzi/ralph-copilot at e5b2813cc876c73a8c9d3398c0115da0d15f63cf · GitHub
Copilot implementation of Ralph loop. Contribute to giocaizzi/ralph-copilot development by creating an account on GitHub.
giocaizzi/ralph-copilot
GitHub
html-meta
giocaizzi/ralph-copilot
2026-08-13T14:24:04
ale-0288
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
Ralph (snarktank)
https://github.com/snarktank/ralph
external
github.com
Ryan Carson's PRD-driven Ralph implementation that re-runs Amp or Claude Code with a fresh instance per iteration, gates each story on typecheck and tests, and persists state in prd.json, progress.txt, and Git history until every story passes.
Ryan Carson's PRD-driven Ralph implementation that re-runs Amp or Claude Code with a fresh instance per iteration, gates each story on typecheck and tests, and persists state in prd.json, progress.txt, and Git history until every story passes.
Ryan Carson's PRD-driven Ralph implementation that re-runs Amp or Claude Code with a fresh instance per iteration, gates each story on typecheck and tests, and persists state in prd.json, progress.txt, and Git history until every story passes.
State persistence is explicit enough for repeated runs and handoff. Ryan Carson's PRD-driven Ralph implementation that re-runs Amp or Claude Code with a fresh instance per iteration, gates each story on typecheck and tests, and persists state in prd.json, progress.txt, and Git history until every story passes.
Use Ralph (snarktank) to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
922
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L922
Build
build
Choose runtimes, tools, and delegation surfaces.
verification;state
builder
agent
direct
source-implementation
A
ok
https://github.com/snarktank/ralph
GitHub - snarktank/ralph: Ralph is an autonomous AI agent loop that runs repeatedly until all PRD items are complete. · GitHub
Ralph is an autonomous AI agent loop that runs repeatedly until all PRD items are complete. - GitHub - snarktank/ralph: Ralph is an autonomous AI agent loop that runs repeatedly until all PRD items are complete.
snarktank/ralph
GitHub
html-meta
snarktank/ralph
2026-08-13T14:24:04
ale-0289
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
ralph-claude-code
https://github.com/frankbria/ralph-claude-code
external
github.com
Loop runner that repeatedly re-executes Claude Code against project requirements, using dual-condition exit detection, rate limiting, and a circuit breaker to decide when the loop should stop.
Loop runner that repeatedly re-executes Claude Code against project requirements, using dual-condition exit detection, rate limiting, and a circuit breaker to decide when the loop should stop.
Loop runner that repeatedly re-executes Claude Code against project requirements, using dual-condition exit detection, rate limiting, and a circuit breaker to decide when the loop should stop.
Uses real automated software-engineering systems as evidence for practical loop architectures. Loop runner that repeatedly re-executes Claude Code against project requirements, using dual-condition exit detection, rate limiting, and a circuit breaker to decide when the loop should stop.
Use ralph-claude-code to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
923
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L923
Build
build
Choose runtimes, tools, and delegation surfaces.
exit
builder
agent
direct
source-implementation
A
ok
https://github.com/frankbria/ralph-claude-code
GitHub - frankbria/ralph-claude-code: Autonomous AI development loop for Claude Code with intelligent exit detection · GitHub
Autonomous AI development loop for Claude Code with intelligent exit detection - frankbria/ralph-claude-code
frankbria/ralph-claude-code
GitHub
html-meta
frankbria/ralph-claude-code
2026-08-13T14:24:04
ale-0290
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
ralph-orchestrator
https://github.com/mikeyobrien/ralph-orchestrator
external
github.com
Multi-backend implementation of the Ralph Wiggum technique that keeps a coding agent looping until task completion, using role-scoped hat personas that coordinate through events, with human-in-the-loop controls and a monitoring dashboard.
Multi-backend implementation of the Ralph Wiggum technique that keeps a coding agent looping until task completion, using role-scoped hat personas that coordinate through events, with human-in-the-loop controls and a monitoring dashboard.
Multi-backend implementation of the Ralph Wiggum technique that keeps a coding agent looping until task completion, using role-scoped hat personas that coordinate through events, with human-in-the-loop controls and a monitoring dashboard.
Orchestration and control flow are made explicit and inspectable. Multi-backend implementation of the Ralph Wiggum technique that keeps a coding agent looping until task completion, using role-scoped hat personas that coordinate through events, with human-in-the-loop controls and a monitoring dashboard.
Use ralph-orchestrator to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
924
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L924
Build
build
Choose runtimes, tools, and delegation surfaces.
delegation;escalation;exit
builder
agent
direct
source-implementation
A
ok
https://github.com/mikeyobrien/ralph-orchestrator
GitHub - mikeyobrien/ralph-orchestrator: An improved implementation of the Ralph Wiggum technique for autonomous AI agent orchestration · GitHub
An improved implementation of the Ralph Wiggum technique for autonomous AI agent orchestration - mikeyobrien/ralph-orchestrator
mikeyobrien/ralph-orchestrator
GitHub
html-meta
mikeyobrien/ralph-orchestrator
2026-08-13T14:24:04
ale-0291
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
ralphex
https://github.com/umputun/ralphex
external
github.com
Extended Ralph loop runner that creates a Git branch per plan, executes tasks in fresh sessions with a commit after each, runs a multi-phase review pipeline with parallel review agents, and archives the completed plan.
Extended Ralph loop runner that creates a Git branch per plan, executes tasks in fresh sessions with a commit after each, runs a multi-phase review pipeline with parallel review agents, and archives the completed plan.
Extended Ralph loop runner that creates a Git branch per plan, executes tasks in fresh sessions with a commit after each, runs a multi-phase review pipeline with parallel review agents, and archives the completed plan.
Uses real automated software-engineering systems as evidence for practical loop architectures. Extended Ralph loop runner that creates a Git branch per plan, executes tasks in fresh sessions with a commit after each, runs a multi-phase review pipeline with parallel review agents, and archives the completed plan.
Use ralphex to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
925
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L925
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
source-implementation
A
ok
https://github.com/umputun/ralphex
GitHub - umputun/ralphex: Extended Ralph loop for autonomous AI-driven plan execution · GitHub
Extended Ralph loop for autonomous AI-driven plan execution - umputun/ralphex
umputun/ralphex
GitHub
html-meta
umputun/ralphex
2026-08-13T14:24:04
ale-0292
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
ralph (iannuttall)
https://github.com/iannuttall/ralph
external
github.com
File-based Ralph-style agent loop that executes one JSON PRD story per iteration with fresh model context, using Git and on-disk state as memory across Claude, Codex, Droid, and OpenCode backends.
File-based Ralph-style agent loop that executes one JSON PRD story per iteration with fresh model context, using Git and on-disk state as memory across Claude, Codex, Droid, and OpenCode backends.
File-based Ralph-style agent loop that executes one JSON PRD story per iteration with fresh model context, using Git and on-disk state as memory across Claude, Codex, Droid, and OpenCode backends.
Persistent memory is treated as an external runtime artifact. File-based Ralph-style agent loop that executes one JSON PRD story per iteration with fresh model context, using Git and on-disk state as memory across Claude, Codex, Droid, and OpenCode backends.
Use ralph (iannuttall) to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
926
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L926
Build
build
Choose runtimes, tools, and delegation surfaces.
context;state
builder
agent
direct
source-implementation
A
ok
https://github.com/iannuttall/ralph
GitHub - iannuttall/ralph: A minimal, file‑based agent loop for autonomous coding. · GitHub
A minimal, file‑based agent loop for autonomous coding. - iannuttall/ralph
iannuttall/ralph
GitHub
html-meta
iannuttall/ralph
2026-08-13T14:24:04
ale-0293
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
ralph-loop-agent
https://github.com/vercel-labs/ralph-loop-agent
external
github.com
Vercel Labs implementation of the Ralph loop for the AI SDK: an outer loop re-runs the agent with verifier feedback until a verifyCompletion check passes or iteration, token, or cost stop conditions trigger.
Vercel Labs implementation of the Ralph loop for the AI SDK: an outer loop re-runs the agent with verifier feedback until a verifyCompletion check passes or iteration, token, or cost stop conditions trigger.
Vercel Labs implementation of the Ralph loop for the AI SDK: an outer loop re-runs the agent with verifier feedback until a verifyCompletion check passes or iteration, token, or cost stop conditions trigger.
Verification is promoted from a final check to a loop-control signal. Vercel Labs implementation of the Ralph loop for the AI SDK: an outer loop re-runs the agent with verifier feedback until a verifyCompletion check passes or iteration, token, or cost stop conditions trigger.
Use ralph-loop-agent to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
927
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L927
Build
build
Choose runtimes, tools, and delegation surfaces.
trigger;budget;exit
builder
agent
direct
source-implementation
A
ok
https://github.com/vercel-labs/ralph-loop-agent
GitHub - vercel-labs/ralph-loop-agent: Continuous Autonomy for the AI SDK · GitHub
Continuous Autonomy for the AI SDK. Contribute to vercel-labs/ralph-loop-agent development by creating an account on GitHub.
vercel-labs/ralph-loop-agent
GitHub
html-meta
vercel-labs/ralph-loop-agent
2026-08-13T14:24:04
ale-0294
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
Open Ralph Wiggum
https://github.com/Th0rgal/open-ralph-wiggum
external
github.com
Agent-agnostic CLI that runs the Ralph Wiggum loop by feeding the same prompt to a fresh agent instance each iteration, with task tracking, live status monitoring, and mid-loop context injection across six coding-agent backends.
Agent-agnostic CLI that runs the Ralph Wiggum loop by feeding the same prompt to a fresh agent instance each iteration, with task tracking, live status monitoring, and mid-loop context injection across six coding-agent backends.
Agent-agnostic CLI that runs the Ralph Wiggum loop by feeding the same prompt to a fresh agent instance each iteration, with task tracking, live status monitoring, and mid-loop context injection across six coding-agent backends.
Context is managed as durable loop state rather than a single prompt payload. Agent-agnostic CLI that runs the Ralph Wiggum loop by feeding the same prompt to a fresh agent instance each iteration, with task tracking, live status monitoring, and mid-loop context injection across six coding-agent backends.
Use Open Ralph Wiggum to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
928
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L928
Build
build
Choose runtimes, tools, and delegation surfaces.
context
builder
agent
direct
source-implementation
A
ok
https://github.com/Th0rgal/open-ralph-wiggum
GitHub - Th0rgal/open-ralph-wiggum: Type `ralph "prompt"` to start open code in a ralph loop. Also supports a prompt file & status check. Open Code, Claude Code, Codex, Copilot · GitHub
Type `ralph "prompt"` to start open code in a ralph loop. Also supports a prompt file & status check. Open Code, Claude Code, Codex, Copilot - Th0rgal/open-ralph-wiggum
Th0rgal/open-ralph-wiggum
GitHub
html-meta
Th0rgal/open-ralph-wiggum
2026-08-13T14:24:04
ale-0295
Coding-Agent Loop Systems
coding-agent-loop-systems
Pattern
🔁
Compound Engineering
https://every.to/guides/compound-engineering
external
every.to
Every's named plan-work-review-compound loop, where each run feeds lessons back into `AGENTS.md`-style memory so the next loop is easier; the self-improving counterpart to Ralph.
Every's named plan-work-review-compound loop, where each run feeds lessons back into `AGENTS.md`-style memory so the next loop is easier; the self-improving counterpart to Ralph.
Every's named plan-work-review-compound loop, where each run feeds lessons back into `AGENTS.md`-style memory so the next loop is easier; the self-improving counterpart to Ralph.
Persistent memory is treated as an external runtime artifact. Every's named plan-work-review-compound loop, where each run feeds lessons back into `AGENTS.md`-style memory so the next loop is easier; the self-improving counterpart to Ralph.
Use Compound Engineering to choose an implementation surface for repeatable agent work.
Operational pattern or playbook; signal comes from reusable loop structure and practical transferability.
medium
README.md
933
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L933
Build
build
Choose runtimes, tools, and delegation surfaces.
context
builder
agent
direct
operational-pattern
B
ok
https://every.to/guides/compound-engineering
Compound Engineering - Every
The AI-native engineering philosophy
every.to
domain-fallback
2026-08-13T14:24:04
ale-0296
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
Gas Town
https://github.com/steveyegge/gastown
external
github.com
Steve Yegge's multi-agent orchestrator that runs 20-30 parallel coding agents with coordinator, worker, and merge-queue roles; the structured-orchestration end of the spectrum that Ralph anchors with bare iteration.
Steve Yegge's multi-agent orchestrator that runs 20-30 parallel coding agents with coordinator, worker, and merge-queue roles; the structured-orchestration end of the spectrum that Ralph anchors with bare iteration.
Steve Yegge's multi-agent orchestrator that runs 20-30 parallel coding agents with coordinator, worker, and merge-queue roles; the structured-orchestration end of the spectrum that Ralph anchors with bare iteration.
The work separates roles across agents, verifiers, or orchestration layers. Steve Yegge's multi-agent orchestrator that runs 20-30 parallel coding agents with coordinator, worker, and merge-queue roles; the structured-orchestration end of the spectrum that Ralph anchors with bare iteration.
Use Gas Town to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
934
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L934
Build
build
Choose runtimes, tools, and delegation surfaces.
intake;delegation
builder
agent
direct
source-implementation
A
ok
https://github.com/gastownhall/gastown
GitHub - gastownhall/gastown: Gas Town - multi-agent workspace manager · GitHub
Gas Town - multi-agent workspace manager. Contribute to gastownhall/gastown development by creating an account on GitHub.
steveyegge/gastown
GitHub
html-meta
steveyegge/gastown
2026-08-13T14:24:04
ale-0297
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
Amp
https://ampcode.com/
external
ampcode.com
Agentic coding tool built around threads, subagents, and an opinionated harness, with an owner's manual that documents loop-style operating practices.
Agentic coding tool built around threads, subagents, and an opinionated harness, with an owner's manual that documents loop-style operating practices.
Agentic coding tool built around threads, subagents, and an opinionated harness, with an owner's manual that documents loop-style operating practices.
The work separates roles across agents, verifiers, or orchestration layers. Agentic coding tool built around threads, subagents, and an opinionated harness, with an owner's manual that documents loop-style operating practices.
Use Amp to choose an implementation surface for repeatable agent work.
Working implementation, framework, runtime, or repository; signal comes from usable code and ecosystem adoption.
high
README.md
935
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L935
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;context;delegation
builder
agent
direct
implementation
A
ok
https://ampcode.com/
Amp
Amp is the frontier agent that lets you wield the full power of leading models.
ampcode.com
domain-fallback
2026-08-13T14:24:04
ale-0298
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
karl
https://github.com/kayoslab/karl
external
github.com
Autonomous multi-agent development loop with planner, reviewer, architect, tester, developer, deployment, and retry phases.
Autonomous multi-agent development loop with planner, reviewer, architect, tester, developer, deployment, and retry phases.
Autonomous multi-agent development loop with planner, reviewer, architect, tester, developer, deployment, and retry phases.
The work separates roles across agents, verifiers, or orchestration layers. Autonomous multi-agent development loop with planner, reviewer, architect, tester, developer, deployment, and retry phases.
Use karl to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
936
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L936
Build
build
Choose runtimes, tools, and delegation surfaces.
delegation;budget
builder
agent
direct
source-implementation
A
ok
https://github.com/kayoslab/karl
GitHub - kayoslab/karl: Autonomous multi-agent development loop · GitHub
Autonomous multi-agent development loop. Contribute to kayoslab/karl development by creating an account on GitHub.
kayoslab/karl
GitHub
html-meta
kayoslab/karl
2026-08-13T14:24:04
ale-0299
Coding-Agent Loop Systems
coding-agent-loop-systems
Pattern
🔁
joelclaw agent-loop skill
https://github.com/joelhooks/joelclaw/blob/main/skills/agent-loop/SKILL.md
external
github.com
Durable Planner-Implementor-Reviewer-Judge coding loops via Inngest events and progress files.
Durable Planner-Implementor-Reviewer-Judge coding loops via Inngest events and progress files.
Durable Planner-Implementor-Reviewer-Judge coding loops via Inngest events and progress files.
Durable execution and replay are treated as first-class loop infrastructure. Durable Planner-Implementor-Reviewer-Judge coding loops via Inngest events and progress files.
Use joelclaw agent-loop skill to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
937
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L937
Build
build
Choose runtimes, tools, and delegation surfaces.
workspace;delegation;verification;state
builder
agent
direct
operational-pattern
B
ok
https://github.com/joelhooks/joelclaw/blob/main/skills/agent-loop/SKILL.md
joelclaw/skills/agent-loop/SKILL.md at main · joelhooks/joelclaw · GitHub
Personal AI operating system — blog, architecture decisions, and the journey from zero to a composable agent system. - joelclaw/skills/agent-loop/SKILL.md at main · joelhooks/joelclaw
joelhooks/joelclaw
GitHub
html-meta
joelhooks/joelclaw
2026-08-13T14:24:04
ale-0300
Coding-Agent Loop Systems
coding-agent-loop-systems
Tool
🧰
ARIS (Auto-Research-In-Sleep)
https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep
external
github.com
Markdown-only skills that run autonomous overnight ML research loops on Claude Code, Codex, or other LLM agents, iterating idea discovery and experiments with cross-model review as the verification gate.
Markdown-only skills that run autonomous overnight ML research loops on Claude Code, Codex, or other LLM agents, iterating idea discovery and experiments with cross-model review as the verification gate.
Markdown-only skills that run autonomous overnight ML research loops on Claude Code, Codex, or other LLM agents, iterating idea discovery and experiments with cross-model review as the verification gate.
Verification is promoted from a final check to a loop-control signal. Markdown-only skills that run autonomous overnight ML research loops on Claude Code, Codex, or other LLM agents, iterating idea discovery and experiments with cross-model review as the verification gate.
Use ARIS (Auto-Research-In-Sleep) to choose an implementation surface for repeatable agent work.
Inspectable GitHub source; popularity is context, not proof of reliability.
medium
README.md
938
https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md#L938
Build
build
Choose runtimes, tools, and delegation surfaces.
intake;verification
builder
agent
direct
source-implementation
A
ok
https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep
GitHub - wanshuiyin/Auto-claude-code-research-in-sleep: ARIS ⚔️ (Auto-Research-In-Sleep) — Lightweight Markdown-only skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation. No framework, no lock-in — works with Claude Code, Codex, OpenClaw, or any LLM agent. · GitHub
ARIS ⚔️ (Auto-Research-In-Sleep) — Lightweight Markdown-only skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation. No framework, no lock-in — works with Claude Code, Codex, OpenClaw, or any LLM agent. - wanshuiyin/Auto-claude-code-research-in-sleep
wanshuiyin/Auto-claude-code-research-in-sleep
GitHub
html-meta
wanshuiyin/Auto-claude-code-research-in-sleep
2026-08-13T14:24:04