wwwhy725/skillgrad
Developer toolsこのページは AI により自動翻訳されています。技術的な詳細は原文をご確認ください。 原文を見る
SkillGrad: Optimizing Agent Skills Like Gradient Descent
概要
SkillGrad treats an agent skill package as a structured parameter and optimizes it through a textual analogue of gradient descent: an executor produces trajectories, a diagnoser turns them into per-task update signals, a momentum agent accumulates recurring patterns across iterations, and a patcher applies a layer-aware edit to the skill. This repo includes a complete SpreadsheetBench Verified integration as a reference pipeline. To run it, place the dataset under data/benchmarks/spreadsheetbench/ following the layout in data/benchmarks/README.md. To use SkillGrad with a new benchmark, implement the corresponding loader and grader interfaces. Three commands run an end-to-end optimization cycle from the repo root. Switch the backbone model via the MODEL environment variable: The default run id is skillgrad_ . Final artifacts land under results/runs/skillgrad_ /: The training loop optimizes whatever initial skill you point --skills-dir at.
README
以下は原文 README の抜粋です。原文を見る。
SkillGrad
Optimizing agent skills like gradient descent.
SkillGrad treats an agent skill package as a structured parameter and optimizes it through a textual analogue of gradient descent: an executor produces trajectories, a diagnoser turns them into per-task update signals, a momentum agent accumulates recurring patterns across iterations, and a patcher applies a layer-aware edit to the skill.
| Paper | arXiv:2605.27760 |
| Code |
The optimization analogy:
| Gradient descent | SkillGrad |
|---|---|
| parameter θ | structured skill package S = (H, B, Q) |
| loss evidence | task outcomes + trajectory traces |
| per-example gradient | per-task textual diagnosis |
| momentum | persistent pattern memory + per-iteration overlay |
| parameter update | layer-aware skill patch |
Repository layout
pipeline/ executor, diagnoser, momentum, patcher, training loop
prompts/ the five agent prompts
runners/ model dispatch, cost ledger, trajectory logger, …
evaluators/ xlsx grader used as the terminal loss signal
data/ dataset loader, split, layout
scripts/ base_traj.sh, train.sh, eval.sh
seeds/xlsx/ bundled initial skill (SKILL.md)
assets/ figures referenced from the README
Installation
pip install -r requirements.txt
cp .env.example .env
# fill in exactly one of the two provider blocks
This repo includes a complete SpreadsheetBench Verified integration as a reference pipeline. To run it, place the dataset under data/benchmarks/spreadsheetbench/ following the layout in data/benchmarks/README.md. To use SkillGrad with a new benchmark, implement the corresponding loader and grader interfaces.
Quickstart
Three commands run an end-to-end optimization cycle from the repo root.
bash scripts/base_traj.sh # 1. collect failure pool
bash scripts/train.sh # 2. train SkillGrad
bash scripts/eval.sh # 3. evaluate on held-out test split
Switch the backbone model via the MODEL environment variable:
MODEL=gpt-4.1 bash scripts/train.sh
The default run id is skillgrad_. Final artifacts land under
results/runs/skillgrad_/:
results/runs/skillgrad_gpt-5.4/
├── config.json
├── train/
│ ├── training_results.json
│ ├── iter_1/, iter_2/, …
│ └── final_skill/xlsx/
└── eval/
├── summary.json
└── /trajectory.json
Initial skill
The training loop optimizes whatever initial skill you point
--skills-dir at. The repository ships with one bundled seed at
seeds/xlsx/SKILL.md and the scripts default to it. To use a different
initialization, lay it out as /xlsx/SKILL.md and pass
SKILLS_DIR= to the scripts.
Configuration
推奨ツール
別のキーワードを試すか、フィルタを外してください。
インストール
npx skillfish add wwwhy725/skillgrad