gaearon/woodshed — explained in plain English
Analysis updated 2026-07-17 · repo last pushed 2026-01-13
Create a skill as plain-text instructions and test it against multiple real-world prompts.
Automatically grade how well Claude followed a skill's instructions using another Claude instance.
Iterate on a customer service reply skill by reviewing saved results and refining instructions.
Review saved logs of prompts, responses, and grades to learn what worked and what didn't.
| gaearon/woodshed | denissergeevitch/chatgpt-portal | justjake/sqlite3-parser-js | |
|---|---|---|---|
| Stars | 65 | 66 | 64 |
| Language | TypeScript | TypeScript | TypeScript |
| Last pushed | 2026-01-13 | — | — |
| Maintenance | Quiet | — | — |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | pm founder | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Experimental alpha-stage tool that can consume significant Claude API tokens.
An experimental tool for building and testing AI skills for Claude: write instructions and test cases, run them, have Claude grade the results, then iterate.
Mainly TypeScript. The stack also includes TypeScript, Claude API.
Quiet — no commits in 6-12 months (last push 2026-01-13).
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly pm founder.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.