Skip to main content

GitHub Repo Completeness

@ aquarius-wing742026.7.10github-repo-completeness

Scans an existing GitHub repo, produces a P0-P3 missing-item checklist, then walks you through each gap one at a time — you approve before it writes anything.

FeaturedAI Engineering

Basic info

Name
GitHub Repo Completeness
Description
GitHub 项目完备性检查与补全引导。扫描已有 repo,生成缺失清单,按优先级逐项引导用户补全。 不直接生成所有内容,而是引导用户决策后再执行。 Always trigger for: repo completeness, 项目完备性, 项目完善, 项目包装, project packaging, open source ready, 开源准备, github setup, 补全项目, project health, project checklist, 项目清单. Also triggers for: "这个项目还缺什么", "帮我完善项目结构", "准备开源". Do NOT trigger when: 写代码 / 修 bug / code review;从零起草需求或 PRD(走 write-spec); scaffold 全新空项目骨架(走脚手架类 skill);发版 / 打包发布版本(走 release); 发布 skill 到公司库(走 tranfu-publish);部署上线(走 coolify-deploy)。

GitHub Repo Completeness

Scans an existing GitHub repo, produces a P0-P3 missing-item checklist, then walks you through each gap one at a time — you approve before it writes anything.

When to use it

Prepare for open source:

I'm getting the repo ready for GitHub or open source. I want the skill to scan first, hand me a checklist, and let me pick what to fix.

Fill specific gaps:

I already know something's missing — README / LICENSE / CI / issue templates — and I want the skill to walk me through the stack-appropriate fix.

Periodic health check:

The project has been running for a while. I want a quick audit to see if the structure is still complete or has drifted to WEAK.

Check multiple repositories:

I'm short on time. I say "all defaults", the skill runs the whole set, and I review at the end.

Not for:

Drafting a PRD from scratch → write-spec; cold-starting a new project's full doc set → project-init-docs; tagging / releasing → release; deploying → coolify-deploy; scaffolding an empty project skeleton → generic scaffolding skills.

What it produces

Guides you item by item — never writes everything in one shot unless you explicitly say "all defaults" — the most counter-intuitive point.

  • P0-P3 checklist table: four columns = priority / item / status (OK / WEAK / MISSING / N/A) / note
  • Guided fills: each item asks your preference first (which license / how deep the CI / which issue templates) before generating
  • Files touched: may add README.md / LICENSE / .gitignore / CONTRIBUTING.md / CHANGELOG.md / .github/workflows/ci.yml / .github/ISSUE_TEMPLATE/*.yml / .github/pull_request_template.md
  • Final rescan: re-runs the scan to confirm the items you picked moved from MISSING / WEAK to OK, then hands you a summary for sign-off
  • Never does: pick a license for you / tag a release / push to remote / write all missing files without approval

Prerequisites & boundaries

Prereq:

cwd must be the git repo root (git rev-parse --show-toplevel returns and matches pwd); outside a repo it stops and asks.

Adjacent skill split:

Action Goes to
Draft a PRD / requirements doc write-spec
Cold-start a full doc set for a new project project-init-docs
Tag / changelog / release release
Deploy coolify-deploy
Publish a skill to the company library tranfu-publish

Not for:

  • Writing code / fixing bugs / code review
  • Scaffolding an empty project skeleton (use scaffolding skills)
  • Editing GitHub remote metadata (description / topics)

Subtle edges:

  • Single-stack repos work best; for monorepo or mixed stacks the skill asks you to name the primary stack rather than guess
  • License is always "ask you, never pick" — if you want it fast, say "just use MIT" upfront rather than expecting a default

Let's Build Together

Follow us and join the community for updates

WeChat community

Scan to join WeChat group

WeChat QR Code