Menu
Run TinyTest tests and report results. Accepts a file, directory, or test name.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Deep-analyze a single PHP function — trace all code paths, find callers, classify bugs and smells, and generate exhaustive TinyTest unit tests. Useful before /test-cover-file when a function is complex.
Project-wide coverage audit — runs phpdbg across all tests, finds every uncovered function, analyzes each one in context with caller data, generates comprehensive tests for all code paths, and produces a structured bug/smell report with operator decisions. Requires phpdbg.
Project-wide bootstrap — scans all PHP source files, registers any without a test file in test_sources.md, and spawns sub-agents to generate initial test files for each. Run this first on a new project before using test-audit.
Achieve full coverage for a single PHP source file. Phase 1 iterates until every function is called. Phase 2 iterates until every reachable statement and branch is exercised. Requires phpdbg.
Diagnose and fix a failing TinyTest test. Reads the error output, compares test vs source, and repairs the test or explains the source bug.
Generate an initial TinyTest test file for a PHP source file. Creates the test file, runs it, fixes failures, and registers it in test_sources.md. Contains the full TinyTest assertion and annotation reference.
| name | test-run |
| description | Run TinyTest tests and report results. Accepts a file, directory, or test name. |
| argument-hint | [file|dir] [-t test_name] |
-f <file>-d <dir>-t <test_name> alongside -ftests/ directory and use -d tests/Find the tinytest path from AGENTS.md or CLAUDE.md in the project root.
php <tinytest_path>/tinytest.php -j -f <file>
If -j is not available (older tinytest), fall back to verbose mode:
php <tinytest_path>/tinytest.php -v -f <file>
From JSON output, report:
From verbose text output, look for:
err — these are failuresexpected [X] got [Y] — the assertion that failedFor each failure:
Report concisely: "X/Y tests passed. Z failures:" followed by a brief summary of each failure.