auditing
Use when reviewing a bundle-plugin for structural issues, version drift, skill quality, workflow integration, or security risks — before releasing, after…
Use when releasing a bundle-plugin, bumping versions, fixing version drift across manifests, setting up version sync infrastructure, updating CHANGELOG, publishing to marketplaces, or checking release readiness
$ npx -y skills add OdradekAI/bundles-forge --skill releasing --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/releasingContext preview
The summary Claude sees to decide when to auto-load this skill.
Use when releasing a bundle-plugin, bumping versions, fixing version drift across manifests, setting up version sync infrastructure, updating CHANGELOG, publishing to marketplaces, or checking release readiness
name: releasing description: "Use when releasing a bundle-plugin, bumping versions, fixing version drift across manifests, setting up version sync infrastructure, updating CHANGELOG, publishing to marketplaces, or checking release readiness" allowed-tools: Bash(bundles-forge bump-version *) Bash(bundles-forge audit-plugin *) Bash(bundles-forge audit-docs *)
Orchestrate the complete release workflow for a bundle-plugin: verify quality, scan for security risks, check documentation consistency, review change coherence, test locally, bump versions, update documentation, and publish to target platforms.
**Core principle:** Release is a checkpoint, not a formality. Every release deserves the full pipeline — even "minor" version bumps can introduce drift or break platform installs. Users should complete all agent, skill, and workflow development before invoking this skill.
**Skill type: Rigid** — follow every step exactly. Releases have no room for improvisation.
For version management infrastructure details (`.version-bump.json` schema, script usage, version setup for new projects), read `references/version-infrastructure.md`. For distribution strategy options, read `references/distribution-strategy.md`.
**Announce at start:** "I'm using the releasing skill to prepare this project for release."
| Context | Path | |---------|------| | User wants to release a version, provides a project directory | **Path 1: Standard release** — run the full pipeline below | | Urgent fix with small changes | **Path 2: Hotfix release** — run the abbreviated pipeline (see Hotfix Releases) | | Project needs version infrastructure for the first time | **Path 3: Version setup** — read `references/version-infrastructure.md` § Version Setup |
0. Prerequisites → 1. Pre-flight Checks → 2. Address Findings → 3. Change Review & Doc Sync → 4. Local Testing → 5. Version Bump → 6. Release Notes → 7. Final Verification → 8. Publish
Verify all conditions before entering the pipeline. Hard requirements block the pipeline; soft requirements trigger warnings.
**Hard requirements (must pass):**
**Soft requirements (warn if missing):**
git status git tag -l v<version> git branch --show-current
If the working tree is dirty, instruct the user to commit all development work first. If the tag already exists, ask the user to choose a different version. If on a non-main branch, warn and ask for confirmation.
Run all automated checks. If any critical issues are found, resolve them before continuing.
bundles-forge bump-version <target-dir> --check bundles-forge audit-docs <target-dir>
**Plugin validation (Claude Code only):** If running in a Claude Code environment, run `claude plugin validate` (or `/plugin validate` in a session) to verify `plugin.json` schema, skill/agent/command frontmatter, and `hooks/hooks.json` validity. Skip this step on other platforms — the inspector agent covers equivalent structural checks.
**Full audit:** Invoke `bundles-forge:auditing` (preferred — includes qualitative assessment via auditor subagent with 10-category scoring). Fallback: `bundles-forge audit-plugin <target-dir>` (automated checks only, no qualitative scoring).
If audit status is FAIL, resolve critical issues before releasing. If security findings are critical, block the release.
Present all findings to the user grouped by severity:
For fixes, invoke `bundles-forge:optimizing` for quality issues.
This step requires AI judgment — it cannot be fully automated.
**Change coherence review:**
Read the diff from the last release tag to HEAD:
git diff $(git describe --tags --abbrev=0)..HEAD --stat git diff $(git describe --tags --abbrev=0)..HEAD
If no prior tags exist, use `git log --oneline` to identify the scope of changes.
Review all changed files for:
| Check | What to Look For | Severity | |-------|-----------------|----------| | Contradictions | File A says "supports 5 platforms", file B says "supports 4 platforms" | Critical | | Redundancy | Two SKILL.md files with large duplicated sections | Warning | | Over-engineering | Complex abstraction for a simple feature, unnecessary indirection layers | Warning | | Missing registrations | New skill added but not in bootstrap routing, AGENTS.md, or README tables | Critical | | Stale references | Renamed skill but old name still used in prose | Critical |
Present findings as a blocking report using the same Critical/Warning/Info severity model as Step 2. Critical items must be resolved before proceeding.
**Documentation update:**
After resolving coherence issues, sync project documentation:
1. **`docs/` directory** — Check each document references accurate skill names, script commands, and architecture descriptions. Fix any outdated content. 2. **`CLAUDE.md`** — Verify: skill count in Directory Layout, skill names in Lifecycle Flow, scripts in Commands section,
A toolkit for building bundle-plugins — AI coding plugins organized around collaborative skill workflows — across Claude Code, Cursor, Codex, OpenCode, Gemini CLI, and OpenClaw.
Use when reviewing a bundle-plugin for structural issues, version drift, skill quality, workflow integration, or security risks — before releasing, after…
Use when writing, completing, improving, or adapting SKILL.md and agents/*.md in a bundle-plugin — integrating external skills, filling scaffolded stubs, or…
Use when planning new bundle-plugins, splitting complex skills, combining skills into bundles, or exploring a vague idea about packaging skills
Use when optimizing a bundle-plugin or single skill — improving descriptions, reducing tokens, fixing audit findings, restructuring workflows, adding skills to…
Use when generating project structure for new bundle-plugins, adding or removing platform support (Claude Code, Cursor, Codex, OpenCode, Gemini CLI, OpenClaw),…
Use when testing a bundle-plugin locally before release — generating dev-marketplace environments, verifying component discovery, running hook smoke tests, and…