php-qa-ci

How to reference commits, issues, PRs, files and lines in GitHub-posted prose: every reference MUST be a fully clickable Markdown link, never bare text. Use when (ALWAYS, before posting): - Writing a `gh issue comment` or `gh pr comment` body - Writing a `gh issue create` or `gh pr create` body/description - Writing PR review comments - Any prose that will be posted to GitHub and mentions a commit hash, another issue/PR, a file path, or a code line Applies to sub-agents too — any agent posting to GitHub follows this.

Defence Before Fix workflow for preventing bug classes through static analysis. Implements the ratcheting pattern: analyse bug -> create PHPStan rule -> TDD -> fix. Use when: - A bug has been found and you want to prevent the entire bug class - "defence before fix", "create a PHPStan rule for this bug" - "detect this pattern with static analysis", "ratchet this bug class" - User wants to encode institutional knowledge as a PHPStan rule This skill orchestrates 4 phases: 1. ANALYSE - Understand the bug pattern 2. DETECT - Create a PHPStan rule to catch the pattern 3. TDD - Write failing tests reproducing specific bugs 4. FIX - Implement fixes, verify rules and tests pass Reference: https://ltscommerce.dev/articles/defence-before-fix-static-analysis

Branch and PR naming convention enforced by php-qa-ci's branchNamePolicy tool. Use when: - "what branches are allowed for PRs?" - "why did branchNamePolicy fail?" - "can I open a PR from plan/00080-foo?" - "how do I extend the branch allow-list for this project?" - User is on a plan/* branch and wants to know how to proceed This skill answers convention questions and points at the authoritative doc in vendor/lts/php-qa-ci/CLAUDE/branch-policy.md.

🔄 PHP-QA-CI TOOL ORCHESTRATOR - Automatic run→fix→run cycling for php-qa-ci tools. **ONLY for php-qa-ci pipeline tools** ({bin}/qa -t toolname) NOT for ad-hoc tool execution outside php-qa-ci. Use when user requests QA tools via php-qa-ci: - "run phpstan", "use stan skills" - "run tests", "run phpunit" - "run rector", "run cs fixer" - "run allStatic", "run allCS" **CRITICAL**: MUST cycle automatically until tool reports clean OR escalation needed. DO NOT stop after one fix to ask "what next?" - KEEP CYCLING. Supports all php-qa-ci tools via {bin}/qa -t {toolname}: - phpstan (static analysis) - has fixer agent - phpunit (tests) - has fixer agent - rector (refactoring) - self-fixing, re-run until clean - fixer (code style) - self-fixing, re-run until clean - infection (mutation testing) - no auto-fix - allStatic/allCs/allTests - meta groups Automatically detects which tools have fixers and invokes them. **PERFORMS PREFLIGHT CHECK**: Invokes docs-conflict-checker agent to ensure project documentation doesn

Run any php-qa-ci tool or the full pipeline via generic agent delegation. Use for tools that lack a specialized runner skill (rector, fixer, infection, phplint, etc.) or to run the full unfiltered {bin}/qa pipeline. Delegates to: - php-qa-ci_qa-tool-runner agent (haiku) for individual tools - php-qa-ci_full-pipeline-runner agent (haiku) for full pipeline Self-fixing tools (rector, fixer) are automatically re-run until stable. Report-only tools (infection, phplint, etc.) run once and report.

Run PHPStan static analysis and fix errors using intelligent agent delegation. Use when user requests to: - Run static analysis - Fix PHPStan errors - Check code quality - Analyze type errors Delegates to runner agent (haiku) for execution and fixer agent (sonnet) for fixes. Automatically cycles between run and fix until analysis passes or human intervention needed.

Run PHPUnit tests and fix failures using intelligent agent delegation. Use when user requests to: - Run tests (full suite, specific path, or single file) - Fix failing tests - Analyze test failures - Check test coverage Delegates to runner agent (haiku) for execution and fixer agent (sonnet) for fixes. Automatically cycles between run and fix until tests pass or human intervention needed.

Analyze existing PHPStan error logs without running analysis. Use when: - User says "fix the phpstan errors" (after manually running PHPStan) - User says "what phpstan errors do I have?" - User points to specific log file - PHPStan was run outside Claude's context Delegates to fixer agent (sonnet) to find logs, parse errors, and implement fixes. Does NOT execute PHPStan - use phpstan-runner for that.

Analyze existing PHPUnit test failure logs without running tests. Use when: - User says "fix the test failures" (after manually running tests) - User says "what tests are failing?" - User points to specific log file - Tests were run outside Claude's context Delegates to fixer agent (sonnet) to find logs, parse failures, and implement fixes. Does NOT execute tests - use phpunit-runner for that.