github-address-comments

📁 kentoshimizu/sw-agent-skills 📅 1 day ago
0
总安装量
1
周安装量
安装命令
npx skills add https://github.com/kentoshimizu/sw-agent-skills --skill github-address-comments

Agent 安装分布

amp 1
cline 1
opencode 1
cursor 1
continue 1
kimi-cli 1

Skill 文档

Github Address Comments

Overview

Use this skill to turn review comments into prioritized, verified fixes and explicit thread closure evidence.

Scope Boundaries

  • Use this skill when the task matches the trigger condition described in description.
  • Do not use this skill when the primary task falls outside this skill’s domain.

Templates And Assets

  • Comment resolution log:
    • assets/comment-resolution-log-template.md

Shared References

  • Reviewer reply patterns:
    • references/reply-templates.md

Inputs To Gather

  • Target PR information.
  • Open review comments and thread status.
  • Repository validation requirements.
  • Scope agreement for this response pass.

Deliverables

  • Prioritized comment-action mapping.
  • Code changes scoped to accepted comment threads.
  • Thread-by-thread responses with verification evidence.
  • Deferred-item log for out-of-scope threads.

Workflow

  1. Confirm gh authentication and identify active PR.
  2. Fetch comments using scripts/fetch_review_threads.py.
  3. Prioritize threads by severity/risk/dependency and log in assets/comment-resolution-log-template.md.
  4. Implement focused fixes and run relevant validation.
  5. Reply with concrete change references using references/reply-templates.md.

Scripts

  • Fetch review threads:
    • python3 scripts/fetch_review_threads.py --repo . --pr <number>
  • JSON output for tooling:
    • python3 scripts/fetch_review_threads.py --repo . --pr <number> --json

Quality Standard

  • Every addressed comment maps to code changes or explicit rationale.
  • High-severity comments are handled before cosmetic threads.
  • Behavior-affecting fixes include verification evidence.
  • Responses are specific enough for quick reviewer validation.

Failure Conditions

  • Stop when comment intent or scope is ambiguous.
  • Stop when requested change conflicts with approved product/architecture decisions.
  • Escalate when required context is missing from reviewer discussion.