Run Tests

SkillDev tools

This skill lets your AI run the tests for your project when you ask. It can run pytest with parallel execution and coverage, and build the correct dotnet test command for a repo with the right filters and options. Use it to check that your changes work before moving on.

Available today. Use it from your connected AI after setup.

After adding the skill, just ask your AI to run tests or verify your changes. Saying run tests, test, or pytest will set it off.

Then ask your AI: use the Run Tests skill

What your AI can do with it

  • Run pytest with parallel execution
  • Report code coverage with test runs
  • Build the correct dotnet test command for a repo
  • Apply the right filters and options to a test run
  • Verify your changes by running the tests

What this skill tells your AI

The instructions your AI receives, as published by agentculture/culture in .claude/skills/run-tests/SKILL.md and read by ahel’s review.

Run the project's pytest suite with optional parallelism (pytest-xdist) and coverage.

Usage

# Default: parallel + verbose (recommended)
bash .claude/skills/run-tests/scripts/test.sh -p

# Quick check: parallel + quiet
bash .claude/skills/run-tests/scripts/test.sh -p -q

# Full CI mode: parallel + coverage + xml report
bash .claude/skills/run-tests/scripts/test.sh --ci

# Specific test file
bash .claude/skills/run-tests/scripts/test.sh -p tests/test_socket_server.py

# Without parallelism (for debugging test ordering issues)
bash .claude/skills/run-tests/scripts/test.sh tests/test_rooms.py

# With coverage
bash .claude/skills/run-tests/scripts/test.sh -p -c

Options

FlagShortDescription
--parallel-pRun with -n auto (pytest-xdist, uses all CPU cores)
--coverage-cEnable coverage reporting to terminal
--ciFull CI mode: parallel + coverage + XML report + verbose
--quick-qQuiet output (no verbose, no coverage)

Extra arguments are passed through to pytest (e.g., -x for stop-on-first-failure, -k "pattern" for filtering).

When to Use Which Mode

  • After code changes: bash .claude/skills/run-tests/scripts/test.sh -p — fast parallel run, verbose output
  • Quick sanity check: bash .claude/skills/run-tests/scripts/test.sh -p -q — minimal output
  • Before PR / release: bash .claude/skills/run-tests/scripts/test.sh --ci — matches CI exactly
  • Debugging flaky test: bash .claude/skills/run-tests/scripts/test.sh tests/test_flaky.py — sequential, single file

Provenance

Canonical supplier is agentculture/guildmaster (post steward→guildmaster cutover). Culture's scripts/test.sh is intentionally ahead of guildmaster's copy: it adds xdist shard-combine, stale-shard wipe, coverage-floor (exit 2) surfacing, and combine/report/xml failure propagation that guildmaster's simpler exec pytest --cov lacks. Do not overwrite from upstream on resync; the improvements are offered back to guildmaster.

Signals

GitHub stars
113
Forks
6
Last commit
Aug 2026

ahel review

  • K6low
    bundled executables the agent is told to run

Automated review, not a security audit. Ruleset v1+k2.

Advanced
Catalog kind
skill
Gateway key
run-tests
Source
github.com/agentculture/culture