pr

📁 seabbs/claude-code-config 📅 3 days ago
1
总安装量
1
周安装量
#47812
全站排名
安装命令
npx skills add https://github.com/seabbs/claude-code-config --skill pr

Agent 安装分布

amp 1
opencode 1
kimi-cli 1
codex 1
gemini-cli 1

Skill 文档

Create a pull request based on issue #$ARGUMENTS. Follow these steps systematically:

Phase 0: Setup and Worktree Decision

  1. Ask the user: “Do you want to use a git worktree for this issue? (y/n)”
    • If yes: Create worktree ./worktree-issue-$ARGUMENTS-[desc] with branch issue-$ARGUMENTS-[desc]
    • If no: Create a new branch: git checkout -b issue-$ARGUMENTS-[brief-description]

Phase 1: Issue Analysis and Complexity Assessment

  1. Analyse the issue:

    • Launch a haiku agent to: fetch issue #$ARGUMENTS with gh issue view, summarise it, identify related code areas using Glob/Grep, and return structured findings
  2. Assess complexity (SIMPLE / MEDIUM / COMPLEX):

    • SIMPLE: Single file, typos, docs-only, trivial test updates
    • MEDIUM: 2-5 files, bug fixes, standard test updates
    • COMPLEX: New features, architectural changes, statistical implementations
  3. Execute workflow based on tier:

FOR SIMPLE ISSUES:

  • Quick codebase search for relevant code
  • Implement changes directly
  • Basic linting check only
  • Skip to Phase 4

FOR MEDIUM ISSUES:

  • Launch parallel haiku agents to: (a) search codebase for relevant files, (b) search CLAUDE.md and project docs for relevant guidelines
  • Create lightweight plan in memory
  • Continue to Phase 2

FOR COMPLEX ISSUES:

  • Write issue analysis to issue_analysis_$ARGUMENTS.md
  • Launch parallel haiku agents for codebase search and context mining
  • Launch a sonnet agent to: create a detailed implementation plan with file modifications, code snippets, and testing strategy
  • Write plan to pr_plan_issue_$ARGUMENTS.md
  • Iterate on plan if needed
  • Continue to Phase 2

Phase 2: Implementation

  1. Implement the changes:
    • For statistical/research code: use /stats-implement patterns
    • Write and run tests
    • Fix based on test results
    • For independent tasks, use parallel subagents

Phase 3: Quality Assurance (complexity-based)

SIMPLE: Basic lint check and run tests if applicable

MEDIUM: Run lint, docs check, code review, tests, and requirements check

COMPLEX: All of MEDIUM plus statistical review (if applicable) and academic writing review (if applicable)

Phase 4: Final Verification and PR Creation

  1. Fix any issues from quality checks
  2. For MEDIUM/COMPLEX: Self-review the completed work
  3. Verify all acceptance criteria are met
  4. For COMPLEX: Clean up planning documents
  5. Create well-structured commits, push the branch, and create PR:
    • Title: “Fix #$ARGUMENTS: [Clear description]”
    • Prepend: **This is entirely from an agent so do not review until I have pinged for review as I will do a first pass**
    • Include summary, changes, testing, and related issues