review-and-release
Skill12 starsChanged 2 months agoMITView on GitHub
Sync README with current agents/skills, validate plugin conventions, sync the Codex plugin directory (plugins/ai-kit), bump version, and commit. Use when the user says "/review-and-release" or wants to prepare a release after adding, removing, or changing agents or skills.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- ivklgn/bag-of-skillsword for word, first seen Sep 4, 2026
Other versions1 project uses a version of this.
Word for word · 1
- ivklgn/bag-of-skills★ 12100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- new-skillSkill
- 12-factor-appsSkill
- ask-meSkill
- audit-websiteSkill
- can-i-useSkill
- code-reviewerSkill
- compatibility-auditSkill
- explain-branch-changesSkill
- humanizer-ruSkill
- humanizerSkill
- jsdocSkill
- load-branch-changesSkill
- nextjs-developerSkill
- recapSkill
- reset-permissionsSkill
- review-golangSkill
- seo-auditSkill
- simplify-code-commentsSkill
- test-health-checkSkill
- update-golang-depsSkill
- update-node-depsSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 9 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
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
- could be automatic: a setting could enforce it
- Distinctive: line 33
Read README.md and parse its three inline lists (backtick-separated names, alphabetical):
- Distinctive: line 45
If zero differences across agents and skills, report "README is already in sync" and skip to Step 3.
- Distinctive: line 49
Check all agent and skill files against these rules:
- Distinctive: line 96
If any issues found, present them as a numbered list with file paths and stop.
- Very specific: line 100
If Step 2 found differences, update README.md using the Edit tool: rebuild each affected inline list (## Agents, ## Skills, ## Commands) …
- Distinctive: line 109
Use the Edit tool on each file.
- Distinctive: line 131
Before committing, show a clear summary:
- Distinctive: line 156
Only show sections with actual changes.
- Distinctive: line 166
Commit with message format:
- Distinctive: line 175
Do NOT use git add -A or git add ..
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Review and Release
Synchronize README.md with the current set of agents and skills, validate files follow plugin conventions, sync the Codex plugin directory, bump the patch version in both plugin manifests, and create a release commit.
Step 1: Inventory Agents and Skills
Scan the filesystem to build the current state.
Agents:
ls agents/*.mdFor each .md file, read the YAML frontmatter and extract name and description. Condense each description into a short README-style summary (under ~90 characters), matching the style of existing entries.
Skills:
ls skills/*/SKILL.mdFor each SKILL.md, read the YAML frontmatter and extract name and description. Condense each description into a short README-style summary.
Store both inventories for use in subsequent steps.
Step 2: Diff Against README
Read README.md and parse its three inline lists (backtick-separated names, alphabetical):
## Agents—`name` · `name` · …## Skills— same format## Commands—`/ai-kit:name` · …(one entry percommands/*.mdfile)
Compare the filesystem inventory against README and classify:
- Added — in filesystem but not in the README list
- Removed — in the README list but not in filesystem
- Unchanged — matches
If zero differences across agents and skills, report "README is already in sync" and skip to Step 3.
Step 3: Validate Plugin Conventions
Check all agent and skill files against these rules:
Agent files (agents/*.md)
- Frontmatter present — starts with
---and has closing--- - Required fields —
name,description,tools,modelall present - Name matches filename —
namefield equals filename without.md - Model value — one of:
haiku,sonnet,opus,inherit - Description is non-empty
- Meaningful markdown body after frontmatter
Codex agent manifests (agents/*.toml)
Codex discovers subagents from .toml files, so every Claude Code agent must have a matching Codex manifest.
- Pairing — every
agents/*.mdhas a siblingagents/*.toml(same basename), and everyagents/*.tomlhas a siblingagents/*.md. Report any unpaired file. - Valid TOML — parses without error.
- Required fields —
name,description,sandbox_mode,developer_instructionsall present and non-empty. - Name matches —
nameequals the filename without.tomland equals thenamein the paired.md. sandbox_modevalue — one of:read-only,workspace-write. Useread-onlywhen the paired.mdtoolscontain noWrite/Edit/Bash, otherwiseworkspace-write.developer_instructionsmatches the paired.mdmarkdown body (the prompt after frontmatter).
Skill files (skills/*/SKILL.md)
- Frontmatter present — starts with
---and has closing--- - Required fields —
nameanddescriptionpresent. Eithermodelordisable-model-invocation: truemust exist. - Name matches directory —
namefield equals parent directory name - Model value (if present) — one of:
haiku,sonnet,opus - Description is non-empty
- Meaningful markdown body after frontmatter
Plugin manifests
Validate both .claude-plugin/plugin.json and .codex-plugin/plugin.json:
- Valid JSON
- Required fields —
name,description,versionpresent - Version matches semver
X.Y.Z - Both manifests have the same
versionvalue
Codex interface
Validate agents/openai.yaml:
- File exists
interface.display_nameis present and non-empty
If any issues found, present them as a numbered list with file paths and stop. Do not proceed until issues are fixed.
Step 4: Apply README Updates
If Step 2 found differences, update README.md using the Edit tool: rebuild each affected inline list (## Agents, ## Skills, ## Commands) sorted alphabetically, keeping the existing `name` · `name` format.
Step 5: Bump Version
Increment the patch version (e.g., 2.0.4 → 2.0.5) in both manifests so Claude Code and Codex stay aligned:
.claude-plugin/plugin.json.codex-plugin/plugin.json
Use the Edit tool on each file.
Step 6: Sync Codex Plugin Directory
The Codex marketplace manifest (.agents/plugins/marketplace.json) points at plugins/ai-kit/, not the repo root — Codex users get ONLY what is inside that directory. It must mirror the root on every release:
rsync -a --delete skills/ plugins/ai-kit/skills/
rsync -a --delete agents/ plugins/ai-kit/agents/
cp .codex-plugin/plugin.json plugins/ai-kit/.codex-plugin/plugin.jsonVerify the sync:
diff -rq skills plugins/ai-kit/skills && diff -rq agents plugins/ai-kit/agentsBoth diffs must be empty. Commands are intentionally not synced — the Codex plugin manifest declares only skills and mcpServers; commands/ is Claude-only.
Step 7: Present Summary
Before committing, show a clear summary:
## Release Summary
**Version:** X.Y.Z → X.Y.Z+1
**README changes:**
- Added agents: ...
- Removed agents: ...
- Added skills: ...
- Removed skills: ...
- Added commands: ...
**Codex sync:** plugins/ai-kit updated (N files changed) / already in sync.
**Validation:** All agents and skills pass convention checks.
**Files to commit:**
- README.md
- .claude-plugin/plugin.json
- .codex-plugin/plugin.json
- plugins/ai-kit/** (if the sync changed anything)Only show sections with actual changes.
Step 8: Commit
Stage and commit only the modified files:
git add README.md .claude-plugin/plugin.json .codex-plugin/plugin.json plugins/ai-kitCommit with message format:
release: vX.Y.Z
- Updated README (added N agents, removed M agents, ...)
- Bumped Claude and Codex plugin versions to X.Y.ZDo NOT use git add -A or git add .. After committing, run git status to confirm clean state and report the commit hash.
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.

