package-boost

Use when prepping package release, `.gitattributes` missing/thin, or keeping `composer archive` / Packagist `--prefer-dist` tarballs lean. Activates: lean dist, gitattributes, export-ignore, dist size, packagist tarball, composer archive bloat, lean-package-validator, stolt lpv, AI files in dist, .ai shipping to dist, .claude in tarball, CLAUDE.md in tarball, AGENTS.md in tarball.

Use when developing Laravel packages with Orchestra Testbench. Activates: writing package code, running package tests, working with testbench, configuring service providers; mentions: package, testbench, package development, service provider.

MUST USE when authoring an AI skill — creating a new SKILL.md, naming a skill, or deciding where one lives. Covers namespacing across host/vendor/shipped skills, auto-activating frontmatter, `.ai/skills/` vs `resources/boost/skills/` choice, and `boost sync` regen. Activates: creating a skill, adding a skill, drafting a SKILL.md, naming a skill, choosing where a skill lives, editing `.ai/skills/**/SKILL.md` or `resources/boost/skills/**/SKILL.md`; mentions: skill, SKILL.md, skill name, skill namespace, skill collision, vendor skill, shipped skill.

Pre-push / pre-release checklist. Local quality gates → docs audit → commit + push → watch CI → draft release notes once CI is green. Activate before: pushing release-bound commits, tagging a release, drafting release notes, or when user mentions: pre-release, pre-push, release checklist, ship, cut release, release notes.

Use when CI matrix has gone red and needs diagnosis. Activates: ci matrix fail, matrix red, prefer-lowest fail, prefer-stable fail, dependency conflict, composer resolve, version excluded, security-advisories floor, testbench phpunit interlock, matrix cell regression; mentions: a specific cell failing in GitHub Actions.

Use when writing/reviewing package code spanning multiple Laravel or PHP majors. Activates: adding dependency, framework API across range, composer constraints, version-guarded code; mentions: composer constraint, illuminate/support version, laravel version matrix, backwards compatibility, version guard, feature detection, minimum laravel version, support older laravel, testbench.

Use when writing or improving README files. Covers required sections, structure, voice, staleness-audit pattern. Laravel-package conventions in `references/laravel-package.md`. Activates: writing a README, improving an existing README, auditing a README for staleness; mentions: readme, README.md, project documentation, package docs.

Use when drafting GitHub release bodies from a tag range or staged changes. Default: GitHub's auto-generated format; override only on major/breaking. Laravel notes in `references/laravel-package.md`. Activates: writing release notes, drafting a release announcement, summarizing a tag range; mentions: release notes, GitHub release, RELEASE_NOTES.

Use when writing UPGRADING.md or UPGRADE.md — migration steps between major versions. Covers per-major sections, before/after code, version-jump rules. Laravel conventions in `references/laravel-package.md`. Activates: writing UPGRADING.md, drafting an upgrade guide, planning a major version bump; mentions: UPGRADING, upgrade guide, migration steps.

MUST USE when editing this repo's AI guidelines or deciding where a new skill belongs. Repo-specific guardrail: never edit auto-generated files; pick the right source dir (`.ai/skills/` internal vs. `resources/boost/skills/` shipped); always re-sync after `.ai/` edits. For skill-authoring mechanics (frontmatter, naming, namespacing), defer to the shipped `skill-authoring` skill. Activates when: editing any file in `.ai/`, editing any guideline under `.ai/guidelines/`, choosing where a new skill lives in this repo, editing any auto-generated guideline / skill dir (`CLAUDE.md`, `AGENTS.md`, `GEMINI.md`, `.claude/skills/`, `.cursor/skills/`, `.agents/skills/`, `.github/skills/`, `.junie/skills/`, `.kiro/skills/`), or user mentions: guidelines, CLAUDE.md, AGENTS.md, GEMINI.md, AI configuration, sync.

Runs backend code quality checks in two tiers: Pint + related tests (every change), PHPStan + full suite (completion only). Activate after making changes to PHP files, or when user mentions: phpstan, pint, code quality, static analysis, code style, run checks.

Test-driven bug fixing workflow. Activates when: fixing bugs, debugging issues, resolving defects, investigating errors, or when user mentions: bug, fix, broken, not working, error, issue, defect, regression.

Requests an independent code review from OpenAI Codex CLI, critically evaluates its findings, and applies warranted fixes. Activates when: the user says /codex-review, asks for a Codex review, or wants an external AI review of changes.

Applies PR review feedback with critical evaluation. Activates when: applying review comments, addressing PR feedback, responding to code review, or when user mentions: review feedback, PR comments, apply feedback, address comments, reviewer feedback.