agentleFS
Sign inStart free
crimeacscrimeacs/auto-improve/SKILL.md

auto-improve

Skill132 starsChanged 5 months agoMITView on GitHub

GAN-style iterative improvement loop for any text artifact. Mutates a file, grades each change against a rubric with a SEPARATE model, keeps only verified wins (pairwise-judged), reverts the rest. The git history is the improvement log. Use when the user wants to autonomously improve the quality of a document, email, prompt, landing page, README, contract, or any text file.

  • Reads credentials

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. Ronnasayd/aipp★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If you don't pass --criteria, auto-improve writes a rubric from the artifact first and saves it to results/<tag>.rubric.md — read it to see …

    line 30
  2. Very specific:

    Check a finished run: python3 improve.py --status --tag v1.

    line 32
  3. Distinctive:

    Never edit outside the artifact.

    line 55

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

auto-improve

Autonomous, judge-gated self-improvement for any text artifact. See README.md for the full guide; this is the agent-facing quick reference.

When to use

The user wants something measurably better, not just rewritten — and is willing to express "better" as a rubric. Good fits: cold emails, landing copy, prompts, skill definitions, blog posts, cover letters, docs, contracts, menus.

Run it

export GEMINI_API_KEY=...        # https://aistudio.google.com/apikey
python3 improve.py \
  --artifact path/to/file.md \   # must be inside a git repo
  --tag v1 \                     # → branch improve/v1, results/v1.tsv
  --max-iterations 8 \
  --criteria path/to/rubric.md   # OPTIONAL — omit to auto-infer a rubric; or add
                                 # --goal "what good looks like" to steer the inferred one

If you don't pass --criteria, auto-improve writes a rubric from the artifact first and saves it to results/<tag>.rubric.md — read it to see what it's optimizing for. Check a finished run: python3 improve.py --status --tag v1. The result lives on the improve/<tag> git branch — git diff main improve/v1 -- <file>.

How it decides (the anti-slop part)

  1. Mutate — best-of-N candidate edits, applied as surgical diffs (crash-proof).
  2. Score — each candidate graded against the rubric by a separate model call.
  3. Decide — the best candidate is pairwise-judged against the champion (both orderings, debiased); kept only if it genuinely wins, else reverted.

The mutator never grades its own work, and confident-but-worse rewrites are discarded — so the climb is real and every commit is a verified gain.

Writing the rubric

A markdown file of weighted dimensions totaling 100 — anchored (50/70/90), reward-framed, specific. See criteria/README.md and the worked examples in criteria/.

Key rules

  • The mutator and evaluator never share context.
  • One artifact, one rubric, one --tag per run.
  • The git branch is the source of truth; results/<tag>.tsv is the climb log (untracked).
  • Never edit outside the artifact; small surgical diffs, not wholesale rewrites.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free