Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
107 lines (81 loc) · 4.65 KB

CONTRIBUTING.md

File metadata and controls

107 lines (81 loc) · 4.65 KB
Raw

Contributing to test-plan

Thanks for contributing! This repo is a Claude Code plugin (test-plan) providing skills for end-to-end test planning for RHOAI — see the README for the skill catalog and CLAUDE.md for the architecture. Working conventions for agents and humans live in AGENTS.md.

Prerequisites

  • Python ≥ 3.10 (CI runs 3.14)
  • uv for dependency management
  • git

Setup

git clone git@github.com:fullsend-ai/test-creator.git
cd test-creator
uv sync --extra dev
uv run pre-commit install

Everyday commands

All build, test, and lint commands live in AGENTS.md — the canonical reference for humans and agents alike. In short: pytest for tests, pre-commit for ruff and flake8, make lint for skillsaw.

Project layout

Path Contents
scripts/ Deterministic Python CLIs (validation, parsing, Jira, repo ops) — no LLM calls
scripts/utils/ Shared utilities (schemas, frontmatter, error handling)
skills/ One SKILL.md per skill; shared scripts via skills/_common/scripts
tests/unit/, tests/integration/ Tests, run with uv run pytest
tests/consts/ Shared test data by domain (<domain>_constants.py)
tests/helpers.py, tests/conftest.py Artifact factories and shared fixtures (run_cli, git_repo, feature_dir)
docs/ Human review guide for generated test plans

Making changes

Follow the rules in AGENTS.md — the short version:

  • Keep changes scoped to the issue's acceptance criteria; prefer targeted edits over new frameworks.
  • Deterministic logic goes in scripts/ with JSON output; SKILL.md files stay instruction-only.
  • New behavior needs tests. Shared fixtures go in tests/consts/, not inside test modules.
  • Line length 120; constant and function names must be unique repo-wide (flake8 UUC/UFN).

Commits and DCO

  • Use conventional commit subjects: fix:, feat:, docs:, chore:, test:, refactor: — imperative mood, referencing the issue (Closes #NN).
  • This repository enforces the Developer Certificate of Origin: sign off your commits with git commit -s. Commits pushed by the fullsend bot identities are exempt (the DCO app skips bots).

Opening a pull request

CI runs on every PR:

  • Lint — skillsaw (strict) and pre-commit on Python 3.14.
  • Tests — full pytest suite with --cov-fail-under=70.
  • DCO — sign-off check.

CodeRabbit also reviews PRs using the per-path instructions and threat model in .coderabbit.yaml.

Working with the fullsend agents

This repository runs fullsend agents:

  • Every PR is auto-reviewed by the review agent when opened or pushed. Outcome labels: ready-for-merge (approved), requires-manual-review (needs human judgment), plus risk/* classification. When the review agent requests changes, the fix agent runs automatically on bot-authored PRs; human-authored PRs need the fullsend-fix label to opt in.

  • Slash commands (as PR/issue comments):

    Command Where Effect
    /fs-triage Issue Re-run triage (clears labels, re-evaluates)
    /fs-code [instruction] Issue Have the code agent implement it
    /fs-review PR Request a fresh review round
    /fs-fix [instruction] PR Run the fix agent, optionally with directions
    /fs-fix-stop PR Disable bot-triggered fix runs for this PR

  • Steering the fix agent: it reads only the review body and the text you write after /fs-fix — it does not see inline review comments or general PR discussion. To have it act on a specific inline finding, quote the relevant text in the instruction (multi-line, up to 10,000 bytes; same-repo #NN references work, external URLs do not).

  • needs-human on a PR means the fix loop is approaching its iteration cap and wants your direction — answer with /fs-fix <instruction> or push commits yourself.

Commands require repository access: triage-level for /fs-triage and /fs-review, write-level for the rest — with one exception: the PR author can always use /fs-fix-stop on their own PR, so you can stop the fix agent touching your branch even without write access.

Filing issues

Open an issue with a description, reproduction steps (or the run/artifact that triggered it), and expected behavior. The triage agent labels new issues; ready-to-code hands them to the code agent, but you can always implement them yourself — agent PRs and human PRs follow the same review process.