sync-tools
Skill78 starsChanged 5 months agoMITView on GitHub
Regenerate the Codex distribution copy of the plugin skills, and the copies of the shipped verify tool, and check them for drift. Run after changing anything under claude-plugins/*/skills. OpenCode, Pi and Cursor read the source tree directly and need no sync.
In this repository4 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- harden-task-fileSkill
- learn-from-sessionSkill
- sync-claude-code-pluginsSkill
The file
We split this skill 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
- Very specific: line 20
Run the first command after any change under claude-plugins/*/skills, then commit the regenerated tree with the change.
≈ “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.
Files
/sync-tools — distribution sync
Every host reads the same skill files. The Claude Code plugins under claude-plugins/ are the source; OpenCode (dist/opencode/plugin/index.js), Pi (repo-root package.json, pi.skills) and Cursor (claude-plugins/*/.cursor-plugin/plugin.json, whose plugin root is the source plugin directory) point straight at claude-plugins/*/skills. Only Codex needs a copy, because it installs a plugin into its own cache, and that copy is generated:
python3 scripts/sync_dist.py # regenerate dist/codex/plugins/*/skills byte-identically from source
python3 scripts/sync_dist.py --check # exit 1 on any drift (tests/test_dist_skill_references.py runs this)The same commands copy the verification tool setup-project ships (setup-project/references/harness/verify and verify.mts) to scripts/ and into the eval scaffolds that embed it, and check those copies too.
Run the first command after any change under claude-plugins/*/skills, then commit the regenerated tree with the change. There is no per-host rewrite: shipped skill text names no host primitive (tests/test_skill_frontmatter.py enforces that), so a copy is a copy.
Hand-maintained packaging files
These are not generated. Edit them when what they describe changes:
| File | What it carries |
dist/codex/plugins/manifest-dev/.codex-plugin/plugin.json, dist/codex/plugins/manifest-dev-tools/.codex-plugin/plugin.json | Codex plugin manifests; version must equal the source plugin's .claude-plugin/plugin.json version |
.agents/plugins/marketplace.json | Codex marketplace registry pointing at the two plugin directories |
dist/opencode/plugin/index.js, dist/opencode/plugin/package.json | OpenCode plugin entry (registers the source skill dirs and slash-command wrappers); version tracks manifest-dev's |
dist/opencode/AGENTS.md | OpenCode instructions file the plugin registers |
dist/pi/prompts/*.md | Pi prompt-template aliases (/do, /auto, /babysit-pr) |
claude-plugins/*/.cursor-plugin/plugin.json, .cursor-plugin/marketplace.json | Cursor plugin manifests and marketplace; version must equal the source plugin's |
dist/codex/README.md, dist/opencode/README.md, dist/pi/README.md, dist/cursor/README.md | Per-host install and update instructions |
Version bumps follow the repo's Versioning rules; this skill does not bump anything.
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
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.