agentleFS
Sign inStart free
brentmercadobrentmercado/flux-skill-evolution/CLAUDE.md

flux-skill-evolution

CLAUDE.md0 starsChanged 6 months agoMITView on GitHub

This file orients Claude Code (and contributors) when working inside this repository. For the project overview, see README.md; for the deep dive, see ARCHITECTURE.md. Flux automatically discovers, refines, and validates Claude Code skills for any domain through iterative failure analysis. It combines EvoSkill's evolutionary loop with XSkill's dual-stream knowledge (skills + experiences), adapted to run with no ground-truth benchmark.

  • In this repositoryOne other file it gives its agents.

The file

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

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

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

    All runtime file paths are absolute, under ~/.claude/

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.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
CLAUDE.md

Flux — Autonomous Skill Evolution Engine

This file orients Claude Code (and contributors) when working inside this repository. For the project overview, see README.md; for the deep dive, see ARCHITECTURE.md.

Flux automatically discovers, refines, and validates Claude Code skills for any domain through iterative failure analysis. It combines EvoSkill's evolutionary loop with XSkill's dual-stream knowledge (skills + experiences), adapted to run with no ground-truth benchmark.

Project Structure

.claude/
  skills/flux/SKILL.md          /flux orchestrator (evolve, domains, inspect)
  agents/
    flux-domain-analyst.md      Generates domain profile + synthetic challenges
    flux-executor.md            Attempts challenges using current skills
    flux-evaluator.md           LLM-as-judge rubric scoring
    flux-proposer.md            Failure analysis → skill proposals
    flux-builder.md             Proposals → SKILL.md files + experiences
install.sh                      Deploys to ~/.claude/ for global use
docs/CITATIONS.md               Research lineage (papers not redistributed)

Install

bash install.sh

Runtime Data Layout (generated when you run the engine)

~/.claude/flux/domains/{domain}/       Evolution state per domain
  profile.json                         Domain profile from analyst
  challenges.json                      Generated challenges with rubrics
  frontier.json                        Current best score + skill list
  history.json                         All proposals + outcomes
  experience_bank.json                 Accumulated experiences
  iterations/{N}/                      Per-iteration data
  staging/                             Skills under validation
  rejected/                            Skills that didn't improve score

~/.claude/skills/{domain}/{skill}/     Evolved skills (globally available)
  SKILL.md                             Standard Claude Code skill

Key Design Decisions

  • No ground truth: synthetic challenges + LLM-as-judge replace benchmark scoring
  • Experiences baked into skills: builder integrates experiences as prose, not citation tags
  • Domain-based organization: evolve once, use everywhere (not project-scoped)
  • Greedy frontier: size 1 for v1 simplicity
  • Resumable: state persisted per iteration; --resume continues from last checkpoint
  • Knowledge versioning: skills carry version_context metadata

Development Rules

  • Skill names: kebab-case, no "claude" or "anthropic" (reserved)
  • Agent tools: restricted to the minimum needed (least privilege)
  • Evolved skills must follow Anthropic's SKILL.md format
  • All runtime file paths are absolute, under ~/.claude/
  • Smoke test: /flux evolve "React 19 frontend development" --iterations 2

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