Skip to content

pr-workflow.md

Latest commit

 

History

History
78 lines (55 loc) · 2.38 KB

pr-workflow.md

File metadata and controls

78 lines (55 loc) · 2.38 KB

PR Workflow

Commit Conventions

Format: <type>(<scope>): <description>

Type When to Use
feat New skill or major new capability in an existing skill
fix Fix incorrect behavior in a skill
docs Documentation changes (docs/, README, AGENTS.md)
refactor Restructure a skill without changing its behavior
test Add or update validation guidance, test instructions, or manual test workflows
chore Repository maintenance, tooling changes
perf Improve skill efficiency, clarity, or token usage without changing behavior

Scope is the skill name without the forge- prefix, or a general area:

feat(setup-project): add monorepo detection
fix(address-pr-feedback): correct GraphQL query for nested threads
docs(architecture): add skill file format reference
refactor(create-issue): simplify label selection logic
chore: update workflow references across all skills

Branch Naming

Format: <type>/<issue-number>-<short-kebab-description>

feat/12-add-setup-project-skill
fix/34-graphql-thread-resolution
docs/56-update-architecture-docs

Pre-Commit Checks

See testing.md — Structural Review for the canonical checklist.

PR Checklist

  • Skill follows the standard section order
  • YAML frontmatter is complete and accurate
  • Bash examples are valid and runnable
  • Cross-skill conventions are consistent, including the shared -- <additional context> syntax where applicable
  • Workflow references updated in relevant skills (if workflow order changed)
  • AGENTS.md updated (if guidance or commands changed)
  • README.md updated (if features or docs table changed)

PR Description

## Summary

<1-2 sentences: what changed and why>

Closes #<issue-number>

## Changes

- <skill or file>: <what changed>

## Test Plan

- [ ] Invoked skill via `/forge-<name>` on a test project
- [ ] Verified generated output is correct
- [ ] Checked cross-skill consistency

Review Process

Skills are prompt templates — review focuses on:

  1. Correctness: Do the bash examples work? Are gh commands valid?
  2. Consistency: Do shared conventions match across all skills?
  3. Completeness: Does the skill handle edge cases mentioned in its Guidelines section?
  4. Clarity: Would a future-compatible agent follow the instructions unambiguously?