PHP Code Quality

This documentation was generated with the assistance of AI. Please report any inaccuracies.

Run every static-analysis and coding-style tool a PHP project actually has wired and report every issue they find, grouped by tool and classified by rule id. It is report-only: it never fixes issues, applies Rector/Pint/PHP-CS-Fixer changes, updates a baseline, or edits tool configuration.

Purpose

iru-php-code-quality gives a single, trustworthy quality pass for PHP projects of any flavor. It detects the flavor (Laravel application, WordPress plugin or theme, PrestaShop module, plain Composer library) from composer.json and file markers, which decides the expected tools and the rule-id prefixes that matter: Pint and Larastan for Laravel, PHPCS with WordPressCS and PHPCompatibilityWP for WordPress, PHP-CS-Fixer and PrestaShop’s PHPStan config for PrestaShop, PHP-CS-Fixer/PHPCS with PHPStan or Psalm for libraries, and Rector when rector.php exists. It verifies that each tool’s binary and configuration really exist before trusting a zero result, so a missing tool is reported as skipped and never as "0 issues". When the caller supplies a baseline (for example iru-php-code-one-task-group capturing one before a task group), only new-versus-baseline issues per rule id are reported.

Purpose group

PHP Development Tooling

Plugin

Shipped by the iru-php plugin of the ai-catalog marketplace (install with /plugin install iru-php@ai-catalog; see Plugins).

Invocation

/iru-php-code-quality
/iru-php-code-quality <path-or-glob>

Inputs

Argument Required Description Default

Argument	Required	Description	Default
`path-or-glob`	No	Files, directories or globs to scope both the tools' run and the report (globs are expanded with `find` first, since the tools take paths).	None — each tool analyses what its own config lists.
Baseline (from the prompt)	No	A prior per-tool, per-rule-id issue count or itemized list from an earlier run, supplied by a calling skill. Reported as a diff, treating tools absent from the baseline as all-new.	None — full report. The project’s own PHPStan/Psalm baseline files are always applied by the tools.

path-or-glob

Files, directories or globs to scope both the tools' run and the report (globs are expanded with find first, since the tools take paths).

None — each tool analyses what its own config lists.

Baseline (from the prompt)

A prior per-tool, per-rule-id issue count or itemized list from an earlier run, supplied by a calling skill. Reported as a diff, treating tools absent from the baseline as all-new.

None — full report. The project’s own PHPStan/Psalm baseline files are always applied by the tools.

Tool invocations, all read-only and writing their machine-readable report under build/iru-php-cq-*: PHPStan --error-format=json (with --memory-limit=1G for Larastan, phpstan-wordpress and the PrestaShop extension, which needs PS_ROOT_DIR), Psalm --output-format=json, Pint --test with a report file, PHP-CS-Fixer check --format=json, PHPCS --report=json, Rector --dry-run --output-format=json. PHP runs on the host when possible, otherwise in the project’s Sail, ddev or wp-env container after docker info succeeds.

Outputs

An overall issue total across tools, the detected flavor and environment, then per tool: its total, the config and baseline in force, and a skipped/not-configured note where applicable.
Per tool, one line per rule-id bucket with its count (PHPStan identifier, Psalm type, fixer names, full PHPCS source, Rector class), and the file:line — message list for scoped runs or on request. Docblock-related rules are grouped so a caller can hand them to iru-php-phpdoc; WordPress security, database and i18n sniffs and Psalm Tainted* issues are listed first.
Config-level problems (stale baseline entries, invalid configuration, memory exhaustion, missing standards) are reported under a separate bucket, never as code issues. Expected-but-missing tools for the flavor are recommendations, not failures.
No raw JSON or diff is pasted, and no file, baseline or configuration is modified.

Execution flow

flowchart TD A["Start /iru-php-code-quality\n(optional path or glob)"] --> B["Step 1: detect flavor\nand environment"] B --> C{"PHP usable and\nvendor installed?"} C -- no --> D["Stop: report missing PHP,\nvendor or no tooling wired"] C -- yes --> E["Verify each tool's binary\nand config is wired"] E --> F["Step 2: run each configured tool\nPHPStan, Psalm, Pint,\nPHP-CS-Fixer, PHPCS, Rector"] F --> G{"Tool failed for a\nnon-code reason?"} G -- yes --> H["Report the tooling cause,\nskip that tool"] G -- no --> I["Parse JSON reports"] H --> I I --> J["Step 3: classify by tool\nand rule id"] J --> K{"Baseline supplied?"} K -- yes --> L["Keep only new-versus-baseline\nissues per rule id"] K -- no --> M["Report totals, flavor,\nper-tool buckets"] L --> M

Dependencies

Invokes

None — iru-php-code-quality is a leaf skill; it only runs the project’s own analysis tools. It mentions iru-php-phpdoc only as the skill a caller can hand docblock findings to.

Invoked by

Code — Step 3 and 7, quality baseline and comparison for PHP plans, via the iru-gate-runner agent.
PHP Code One Task Group — Step 1 and 3.7, baseline and regression check, via the iru-gate-runner agent.

Related agents

None — it spawns no agents itself. Calling skills normally run it inside an iru-gate-runner sub-agent.

Source

SKILL.md on GitHub — the file this page was generated from.