agentleFS
Sign inStart free
RuneStone-TechRuneStone-Tech/optimize-anything/AGENTS.md

optimize-anything

AGENTS.md1 starsChanged 8 days agoMITView on GitHub

Instructions for AI coding agents working in this repository. Do not load or read every skill. This repo has 60+ focused optimizers. Read the chosen skill's SKILL.md and follow it: Domains: web, web/_components/skills (focused libraries), styles, motion, backend, architecture, devops. Claude Code: Cursor / VS Code + Copilot: clone/open the repo; manifests are discovered automatically.

The file

We split this AGENTS.md into the rules it gives an agent

9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not load or read every skill.

    line 7
  2. Distinctive:

    If the user named a stack, open that skill directly.

    line 9
  3. Very specific:

    If the request is ambiguous, read skills/router/SKILL.md.

    line 10
  4. Very specific:

    Architecture: only when the problem is boundaries/data/distribution.

    line 25
  5. Very specific:

    Read the chosen skill's SKILL.md and follow it:

    line 37
  6. Distinctive:

    Do not commit unless the user asks.

    line 51
  7. Very specific:

    Do not run destructive commands (terraform apply, kubectl apply, docker push, database migrations) without explicit approval.

    line 52
  8. Very specific:

    Follow CONTRIBUTING.md and .agent/optimizer-skill-builder/SKILL.md.

    line 71
  9. Distinctive:

    Update the three plugin manifests and root README.

    line 72

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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
AGENTS.md

AGENTS.md

Instructions for AI coding agents working in this repository.

Start here

Do not load or read every skill. This repo has 60+ focused optimizers.

  1. If the user named a stack, open that skill directly.
  2. If the request is ambiguous, read skills/router/SKILL.md.
  3. Search the catalog:
python tools/catalog.py search "slow nextjs mui table"
python tools/catalog.py search "postgres connection pool" --limit 5
python tools/catalog.py list --domain backend
  1. Browse skills/CATALOG.md only when search fails.

Select at most 3 skills

  • Primary: framework/runtime/language.
  • Secondary: component library, styling system, motion/3D or infra that is actually involved.
  • Architecture: only when the problem is boundaries/data/distribution.

Examples:

  • Slow Next.js + MUI table → skills/web/nextjs + skills/web/_components/skills/mui.
  • Vue + Tailwind → skills/web/vue + skills/styles/tailwindcss.
  • Rails N+1 → skills/backend/ruby.
  • Docker image size → skills/devops/docker.
  • Monolith vs services → skills/architecture.

Workflow for an optimizer

Read the chosen skill's SKILL.md and follow it:

  1. Scope — detect versions, build, config and the affected route/service.
  2. Audit — run <skill>/scripts/audit.py <project-root> [--json], then confirm by reading code.
  3. Report — use <skill>/assets/report-template.md; findings ranked by impact with evidence.
  4. Plan — quick wins / structural / heavy; stop for approval.
  5. Fix — only approved items, small groups.
  6. Verify — build, test and re-measure the same scenario.

Rules:

  • Evidence, approval, then edits. Do not rewrite before the report.
  • Measure, never guess. Dev mode and a single request are not production evidence.
  • Mark findings Confirmed vs Suspected.
  • Do not commit unless the user asks.
  • Do not run destructive commands (terraform apply, kubectl apply, docker push, database migrations) without explicit approval.

Repo map

skills/<domain>/<skill>/   SKILL.md, README, LICENSE, references/, scripts/audit.py, assets/
skills/router/             entry point for ambiguous requests
skills/CATALOG.md          generated, searchable catalog
skills/catalog.json        generated machine-readable catalog
tools/validate.py          metadata/files/catalog/link validation
tools/catalog.py           catalog generation + search
tools/test_audits.py       scanner contract tests
.agent/optimizer-skill-builder/  how to add a skill

Domains: web, web/_components/skills (focused libraries), styles, motion, backend, architecture, devops.

Add or change a skill

  1. Follow CONTRIBUTING.md and .agent/optimizer-skill-builder/SKILL.md.
  2. Update the three plugin manifests and root README.
  3. Run:
python tools/catalog.py
python tools/validate.py
python tools/test_audits.py

Install

Claude Code:

/plugin marketplace add RuneStone-Tech/optimize-anything
/plugin install optimize-anything

Cursor / VS Code + Copilot: clone/open the repo; manifests are discovered automatically.

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