Communitygithub.com

gener8v/gener8v.claude-skills.engineering

Find and run a project's test, lint, type-check, build and dev-server commands, read the results correctly, and seed a throwaway database when UI work needs realistic data. Never seeds the project's real or default database. Use before 'run the tests', 'lint this', 'does it build', 'start the app so I can see it', 'seed some demo data', and before you claim that a change works. Not for deployment or CI pipeline design; use the project's own deploy tooling for those.

What is gener8v.claude-skills.engineering?

gener8v.claude-skills.engineering is a Claude Code agent skill that find and run a project's test, lint, type-check, build and dev-server commands, read the results correctly, and seed a throwaway database when UI work needs realistic data. Never seeds the project's real or default database. Use before 'run the tests', 'lint this', 'does it build', 'start the app so I can see it', 'seed some demo data', and before you claim that a change works. Not for deployment or CI pipeline design; use the project's own deploy tooling for those.

Works with✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/gener8v/gener8v.claude-skills.engineering/tree/HEAD/skills/development-commands

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Development commands

1. Find the commands

Use the first source that answers. State the command before you run it.

  1. Repository instructions: CLAUDE.md, CONTRIBUTING.md, README.md. In a gener8v pipeline repository, .gener8v/context.md lists the verify commands for each repository.
  2. Task runner: Makefile, justfile, Taskfile.yml, package.json scripts, pyproject.toml tool sections, tox.ini, noxfile.py.
  3. CI workflow: .github/workflows/*.yml. What CI runs defines "the tests pass".
  4. Ecosystem default: resources/ecosystems.md.

If the project runs in containers (compose.yaml, docker-compose.yml, a dev container), run the commands inside the container the way the project documents it. See resources/ecosystems.md.

2. Run

  • Start narrow: one test file or one package. Then run the full suite before you report.
  • Do not start watch mode. Many test runners watch by default. Use the "run once" form (resources/ecosystems.md).
  • Keep the exit code when you shorten output:
    set -o pipefail
    <command> 2>&1 | tail -n 60
    echo "exit=$?"
    
  • Start a dev server in the background. Wait until its URL answers. Stop it when you are done.

3. Read the result

  • The exit code decides. Text such as "done" does not mean success.
  • Count. Report passed, failed and skipped. Skipped, expected-to-fail and "no tests collected" are not passes.
  • Flaky tests. Run a failed test once more, alone. If it passes, report it as flaky. Do not hide it.
  • Failures that were there before. Check the base branch: the latest CI run (gh run list --branch <base> --limit 3) or a run in a separate worktree. Report these failures as "already failing on <base>". Do not use them to excuse your own failures.
  • Do not weaken a test to make it pass. If the test is wrong, say why and change it in a separate, visible commit.

4. Data for UI work

When the UI needs realistic data, follow resources/throwaway-data.md. The short rule: seed only a database that you created for this purpose. Never seed the project's real, default or shared database.

5. Report

Give the command, the exit code, the counts, and the first real error for each failure. If you did not run something (for example, end-to-end tests that need credentials), say so.

Related skills

  • engineering:github-pr-merge and engineering:github-issue-workflow call this skill before a PR.
  • engineering:documentation-bootstrap uses it to find the commands that documents must list.

Related Skills