Claude Code Plugins

Community-maintained marketplace

Feedback

tests-and-formatting-discipline

@diey/codex-agent-skills
0
0

Maintain testing and formatting discipline aligned with the repository CI pipeline without automatically running tools.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name tests-and-formatting-discipline
description Maintain testing and formatting discipline aligned with the repository CI pipeline without automatically running tools.

Tests & Formatting Discipline

Goal

Maintain code quality and CI stability by:

  • Designing changes to be testable
  • Updating or adding tests when behaviour changes
  • Writing code that conforms to formatting standards
  • Never running tests, linters, or formatters without explicit user approval

Project Commands (Reference Only)

Use these exact command names when suggesting verification steps:

  • composer test — Clears cached config and executes the full Pest test suite.
  • composer clean — Runs Laravel Pint and Prettier.
  • composer analyse — Runs Larastan static analysis.

Do not claim these were run unless the user confirms and provides results.

Testing Expectations

  • Behaviour changes must be covered by tests.
  • Prefer feature tests unless logic is truly isolated.
  • Use existing factories, states, and helpers.
  • Follow existing test structure and naming conventions.

If behaviour changes and no test is added:

  • Explicitly state why
  • Offer test coverage as the next step

Formatting Expectations

  • Write code that already conforms to Pint and Prettier standards.
  • Avoid unrelated or drive-by formatting changes.
  • Do not reformat files outside the scope of the change.

Note:

  • On PRs targeting development, the lint workflow may run composer clean and auto-commit formatting fixes.

Tool Execution Policy

  • Never run composer test, composer clean, composer analyse, or any tooling automatically.
  • You may suggest commands, but execution is always the user’s responsibility.

End-of-Task Output

After completing an implementation, provide:

  1. A brief list of files changed
  2. A minimal list of suggested commands the user can run to validate the change

Anti-patterns

  • Running tools without permission
  • Claiming tests, lint, or analysis passed without evidence
  • Changing behaviour without mentioning test coverage
  • Introducing unrelated formatting diffs