Back to the catalog

agent-ready-eval

Evaluate a codebase for agent-friendliness based on autonomous agent best practices. Use when asked to "evaluate for agents", "check agent r

Open source Repository Open in the app JSON README (API)

About

Evaluate a codebase for agent-friendliness based on autonomous agent best practices. Use when asked to "evaluate for agents", "check agent readiness", "audit for autonomous execution", "assess agent-f

Details

Kind
Agent skills
Topic
No topic detected
Publisher
petekp
Origin
majiayu
Category
ferramentas
Stars
11
Forks
11
Last push
2026-09-04T05:53:29Z
Repository state
ativo
Language
Shell
License
MIT
Added
2026-09-02 18:17:57
Updated
2026-09-02 18:17:57
Origin id
petekp/claude-code-setup/skills/agent-ready-eval/SKILL.md@main

README

# Claude Code + Codex Setup

Portable, version-controlled configuration for **Claude Code** and **OpenAI Codex**. Fork it, customize it, sync it across machines.

## Quick Start

```bash
git clone https://github.com/petekp/claude-code-setup.git
cd claude-code-setup && ./setup.sh
```

That's it. Both tools now use this repo's skills and scripts.

## How It Works

The setup script symlinks this repo into `~/.claude/`:

```
~/.claude/
├── skills/    → <repo>/skills/
├── scripts/   → <repo>/scripts/
├── settings.json → <repo>/settings.json
├── statusline-command.sh ← <repo> (copied)
│
~/.mcp.json    → <repo>/.mcp.json
```

Edit files in either location — they're the same files. Commit and push to sync across machines.

### Which Skills Are In This Repo

`skills/` mixes directories authored here with symlinks to skills installed
elsewhere, mostly by [skills.sh](https://skills.sh) into `~/.agents/skills`.

Only the real directories are committed. A symlink stores just its target, and
those targets are absolute paths under one machine's home directory — committing
them would give anyone cloning this repo a dangling link to a directory they
don't have. The linked skills and their sources are recorded in
[SKILLS.md](SKILLS.md) instead, so the set can be rebuilt elsewhere.

Run `./scripts/gen-skills-manifest.sh` after adding or removing a skill to
regenerate `SKILLS.md` and `skills/.gitignore`.

### Codex Skill Syncing

Skills are symlinked individually into `~/.codex/skills/` (Codex doesn't support directory-level symlinks) by `scripts/sync-codex-skills.sh`, which runs automatically on every Claude Code session start. Add a name to `codex-exclude` to skip a skill that conflicts with a Codex built-in or shows up as a duplicate; the file is currently empty, so everything syncs.

### MCP Servers

Claude Code reads MCP servers from `~/.mcp.json` (symlinked). Codex reads them from `~/.codex/config.toml` (manually managed).

## Fork and Customize

1. Fork this repo on GitHub
2. Clone your fork and run `./setup.sh`
3. See [FORKING.md](FORKING.md) for customization guidance

## Syncing

```bash
# Push changes
git add -A && git commit -m "Update config" && git push

# Pull on another machine
git pull && ./setup.sh
```

## Undo

```bash
./setup.sh --undo
```

Removes symlinks and restores backed-up directories.

## Reference Files

Some files aren't symlinked — they're templates or references for manual setup. See [templates/README.md](templates/README.md) for details.

## Troubleshooting

**Skills/commands not appearing in Claude Code?**
1. Check symlinks: `ls -la ~/.claude/skills`
2. Validate frontmatter: `./scripts/validate.sh`
3. Restart Claude Code

**Skills not appearing in Codex?**
1. Check symlinks: `ls -la ~/.codex/skills`
2. Verify the skill isn't in `codex-exclude`
3. Start a Claude Code session (the sync hook runs on SessionStart), or run `./scripts/sync-codex-skills.sh` directly

**Permission denied?**
```bash
chmod +x setup.sh && ./setup.sh
```

**Preview before running?**
```bash
./setup.sh --dry-run
```

**Windows?**
Use WSL or set up symlinks manually.

## License

MIT

More