agentleFS
PalmDrPalmDr/claude-evolving-skills/AGENTS.md

claude-evolving-skills

AGENTS.md8 starsChanged 8 months agoMITView on GitHub

For AI coding agents navigating this repo. Humans: read README.md instead. You're an AI agent reading this. Good. Don't just copy these files — understand the patterns and build your own. Read the SKILL.md files below, understand how they work (parallel agent launches, multi-voice debate, dual-channel scoring, evolution tracking), and create skills adapted to your user's workflow. The goal is for you to evolve your own config, not to clone someone else's. If your user wants a direct install, cp…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 32,671 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository5 other files it gives its agents.

The file

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

8 rules. About 0 are common, 4 are distinctive, and 4 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.

  • 4

    Very specific

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

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

    If your user wants a direct install, cp -r skills/* ~/.claude/skills/ works.

    line 9
  2. Very specific:

    Use WebSearch / WebFetch for web access

    line 49
  3. Very specific:

    Read/write any file in ~/.claude/

    line 50
  4. Distinctive:

    Commit changes to git

    line 52
  5. Distinctive:

    Keep skill files under 600 lines

    line 72
  6. Distinctive:

    Add dependencies beyond the three CLIs (claude, gemini, codex)

    line 76
  7. Distinctive:

    Put personal information in skill files (they're meant to be shared)

    line 78
  8. Very specific:

    Create nested skill directories — one skill = one folder with one SKILL.md

    line 79

≈ “Common” is an estimate: at least 1 in 5 of the 41,581 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

For AI coding agents navigating this repo. Humans: read README.md instead.

How To Use This Repo

You're an AI agent reading this. Good. Don't just copy these files — understand the patterns and build your own. Read the SKILL.md files below, understand how they work (parallel agent launches, multi-voice debate, dual-channel scoring, evolution tracking), and create skills adapted to your user's workflow. The goal is for you to evolve your own config, not to clone someone else's.

If your user wants a direct install, cp -r skills/* ~/.claude/skills/ works. But you'll serve them better by reading these patterns and proposing a tailored setup.

What This Repo Is

A collection of self-evolving skills for Claude Code (~/.claude/skills/). Each skill is a Markdown file with execution instructions that Claude Code follows autonomously — launching parallel agents, searching the web, reading/writing files, and invoking other skills.

The system forms a closed loop: skills scan for improvements → propose changes → debate with multiple models → auto-adopt safe changes → verify next week.

Repo Structure

skills/                     ← The core deliverable. Each subfolder = one skill.
  agentic-radar/SKILL.md    ← 5-agent GitHub/HN/Reddit scanner
  vendor-docs-radar/SKILL.md ← 3-agent official docs monitor
  reflect-and-learn/SKILL.md ← 6-agent self-reflection loop (the most complex skill)
  gemini-agent/SKILL.md      ← Delegates prompts to Gemini CLI
  codex-agent/SKILL.md       ← Delegates prompts to Codex CLI
scripts/                    ← Bash wrappers for cron/launchd automation

How Skills Work

A skill is a single SKILL.md file with YAML frontmatter + Markdown execution instructions:

---
name: skill-name
description: |
  What the skill does and when to trigger it.
---
# Skill Title
## Execution Instructions
### Step 1: ...
### Step 2: Launch N parallel agents
### Step 3: Synthesize
### Step 4: Adopt changes

Claude Code reads the SKILL.md and follows the steps. Skills can:

  • Launch parallel background agents (run_in_background: true)
  • Use WebSearch / WebFetch for web access
  • Read/write any file in ~/.claude/
  • Invoke other skills (e.g., reflect-and-learn invokes gemini-agent and codex-agent)
  • Commit changes to git

Installation Target

Users copy skills/* into ~/.claude/skills/. No build step. Dependencies: Claude Code CLI, git, jq (and optionally Gemini CLI + Codex CLI for multi-model skills).

Key Patterns

Multi-agent parallel launch: Most skills fire 3-6 agents simultaneously, each writing output to a temp working directory, then a synthesis agent reads all outputs.

Adoption classification: Changes are scored and classified as P0-AUTO (safe, auto-apply), P1-AUTO (moderate, auto-apply), HIGH-IMPACT (flag for user), or DEFER (log only).

Multi-voice debate: reflect-and-learn sends proposals to Claude + Gemini + Codex, requires 2/3 agreement for auto-adoption.

Evolution tracking: ~/.claude/evolution-history/evolution-tree.jsonl stores a tree of config changes with parent→child lineage, scores, and retrospective verdicts. ~/.claude/evolution-history/scoreboard.jsonl logs individual change entries with per-change scores and adoption status.

If You're Modifying This Repo

  • Each skill must be self-contained in one SKILL.md file
  • Skills should not assume any specific project directory — they operate on ~/.claude/
  • Keep skill files under 600 lines
  • Scripts in scripts/ are thin wrappers that call claude -p with a skill invocation prompt

Do Not

  • Add dependencies beyond the three CLIs (claude, gemini, codex)
  • Hardcode absolute paths — use ~/.claude/ or relative paths
  • Put personal information in skill files (they're meant to be shared)
  • Create nested skill directories — one skill = one folder with one SKILL.md

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.