Skip to content

AGENTS.md

Latest commit

 

History

History
118 lines (75 loc) · 3.78 KB

AGENTS.md

File metadata and controls

118 lines (75 loc) · 3.78 KB

AGENTS.md — OpenAI Codex Workflow Guide

This file is read by OpenAI Codex (and compatible agents) to understand how to work in this repository. Canonical skill definitions live in .ai/skills/.

Project

WCC (Women Coding Community) Backend — Spring Boot 3.2.5, Java 21, PostgreSQL.

See CLAUDE.md for full architecture, build commands, and conventions.

Skill Access

Repository-specific AI workflows live under .ai/skills/ as canonical markdown runbooks. When a user request matches one of the skills below, read the corresponding file in .ai/skills/ and follow that runbook.

Discovery pattern for Codex:

  • Match the request to a skill name or workflow intent
  • Open .ai/skills/<skill-name>.md
  • Execute the runbook while still respecting this file's repository rules

Skills

commit

Safe commit with sensitive-data detection and conventional commit messages.

Full runbook: .ai/skills/commit.md

Quick reference:

  • Scan diff for secrets before staging anything
  • Conventional Commits format: feat:, fix:, docs:, refactor:, test:, chore:
  • Always write a body explaining why, not just what
  • Stage by file name, never git add .

pr-review

Inline code review posted via GitHub CLI.

Full runbook: .ai/skills/pr-review.md

Quick reference:

  • Use gh pr diff and gh api for inline comments
  • Prioritise: regressions → security → data integrity → conventions → tests
  • Post comments on exact changed lines, not just a summary
  • Java 21 idioms, Spring Boot conventions, Given-When-Then test names

pre-commit-review

Local review of staged and unstaged changes before committing.

Full runbook: .ai/skills/pre-commit-review.md

Quick reference:

  • Review local diff before commit with regression-first mindset
  • Auto-fix style issues in changed files where the runbook allows it
  • Report only findings that require human judgement
  • Enforce Java test naming and @DisplayName conventions

open-pr

Prepare a pull request title and description from the current branch diff.

Full runbook: .ai/skills/open-pr.md

Quick reference:

  • Derive PR title from the dominant conventional-commit change type
  • Fill the repo PR template with only applicable sections
  • Include Swagger screenshots for backend API changes
  • Print the resulting GitHub PR URL for the user

pull-request

Pull request workflow runbook for repository PR creation and description.

Full runbook: .ai/skills/pull-request.md

Quick reference:

  • Use when the user asks for pull request preparation workflows
  • Follow the canonical markdown runbook under .ai/skills/
  • Keep PR output aligned with repository templates and conventions

split-commits

Split local changes into atomic commits grouped by logical concern.

Full runbook: .ai/skills/split-commits.md

Quick reference:

  • Analyse the full uncommitted diff and group files by concern
  • Keep feature, refactor, test, and chore changes separate where possible
  • Re-scan each proposed commit for sensitive data before committing
  • Suggest which commit groups should become separate PRs

Build commands

./gradlew test                  # unit tests
./gradlew testIntegration       # integration tests (Docker required)
./gradlew check                 # all checks including coverage (min 70%)
./gradlew pmdMain pmdTest       # static analysis

Important conventions

  • No AI co-author lines in commit messages
  • @DisplayName with Given-When-Then format for all Java tests
  • AssertJ (assertThat) for all assertions
  • Constructor injection, never field injection
  • No @Value in service fields — use @ConfigurationProperties for groups