Runs after you ship a feature or hit a milestone to sync your code changes across agent memory, CLAUDE.md, and docs so nothing goes stale between sessions. It's built around the idea that docs and memory are the only bridge across sessions and agents, so it enforces three-tier reconciliation: agent memory for your own context, project root markdown for AI guardrails and conventions, and docs for humans and downstream teams. The checklist prevents the two most common failures: bloating CLAUDE.md with changelog narratives instead of rules, and forgetting to update integration guides when you add APIs. Trigger it with "sync up", "tidy up docs", or when you want a clean handoff. Cross platform, works with Claude Code, Codex, OpenCode, and OpenClaw.

skills.sh installs
3.5K
GitHub stars
22.1K
last updated
Aug 21, 2026

Install

$ npx -y skills add kkkkhazix/khazix-skills --skill neat-freak --agent claude-code

Installs into .claude/skills of the current project.

Files

洁癖 — Knowledge and Governance Closeout

你是知识库编辑、规范审计员和收尾者。目标不是「多写一点」,而是让代码、真实运行态、项目文档、Agent 规则、获准维护的记忆和工作区状态彼此一致,让下一次会话或第一次接手的人能找到唯一现役答案。

完成合同

一次洁癖收尾只有在相关事实面都得到明确状态后才算完成:

事实面要回答的问题常见证据
代码现在真正实现了什么?当前分支、schema、配置、测试
运行态用户实际得到什么?deploy marker、服务、真实页面/API、控制台
文档人和下游看到的是不是现役答案?README、架构、接入、运维文档
规则Agent 收到的约束是否同源、可执行、无死引用?层级 CLAUDE.md/AGENTS.md、override、hooks
记忆快照是否仍准确且允许修改?平台记忆入口、索引、生成来源
工作区是否仍有未集成或未审计的残留?会话残留文件、worktree、分支、临时库

每一面标成 verified-current、changed-and-verified、pending、out-of-scope 或 not-applicable。小项目不必硬凑六个面:没有部署就没有运行态面,没有记忆系统就没有记忆面——如实标 not-applicable,不要编造证据。不要把 git status 干净、PR 已合并或测试通过单独当成「全部同步」。发布状态必须区分 draft、PR、merged、deployed、live verified、knowledge closed 和 cleaned。

权限和范围先于洁癖

当前系统、用户和项目规则始终高于本 skill。洁癖扩大检查深度,不扩大操作权限。

先判断请求属于哪一档:

  1. 文档同步:当前项目的代码/文档/规则一致性;记忆默认只读,除非用户或项目收尾规则明确授权写入。
  2. 知识收尾:文档、规则、获准维护的记忆和会话复盘。
  3. 发布收尾:在知识收尾之外核对本地、远端、生产和 live surface;知识凭证完成后才能清场。
  4. 工作区审计:只有用户明确说「整个 workspace / 全部项目 / 审全部」时,才逐项目扩大内容审计。

Sponsors

Details

Repository
github.com/kkkkhazix/khazix-skills
Categories
Git & Pull Requests, Documentation, Code Review & Quality, AI & Agent Building, Release Management, Debugging
First seen
May 16, 2026

More from kkkkhazix/khazix-skills

All 8
1actionbook avatarunsafe-checkeractionbook/rust-skillsCRITICAL: Use for unsafe Rust code review and FFI. Triggers on: unsafe, raw pointer, FFI, extern, transmute, *mut, *const, union, #[repr(C)], libc, std::ffi, Ma3.5Kskills.sh installs
2himself65 avatarearnings-recaphimself65/finance-skillsGenerate a post-earnings analysis for any stock using Yahoo Finance data. Use when the user wants to review what happened after earnings, understand beat/miss r3.4Kskills.sh installs
3getsentry avatarblog-writing-guidegetsentry/skillsWrite, review, and improve blog posts for the Sentry engineering blog following Sentry's specific writing standards, voice, and quality bar. Use this skill when3.4Kskills.sh installs
4himself65 avatarearnings-previewhimself65/finance-skillsGenerate a pre-earnings briefing for any stock using Yahoo Finance data. Use this skill whenever the user wants to prepare for an upcoming earnings report, unde3.4Kskills.sh installs
5dpearson2699 avatarphotokitdpearson2699/swift-ios-skillsImplement, review, or improve photo picking, camera capture, and media handling in iOS apps using PhotoKit and AVFoundation. Use when working with PhotosPicker,3.4Kskills.sh installs
6eronred avataraso-auditeronred/aso-skillsWhen the user wants a full ASO health audit, review their App Store listing quality, or diagnose why their app isn't ranking. Also use when the user mentions "A3.4Kskills.sh installs