Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Turns engineering principles into executable rules for AI coding agents, enabling structured bug fixes, disciplined pull request reviews, safe refactoring, test-first development, and release auditing through a set of skills and commands.
npx claudepluginhub bravekingzhang/agent-coding-playbook --plugin agent-coding-playbookUse when the user asks to fix a bug, investigate an error, debug a crash, explain a stack trace, or make a failing test pass. Enforces reproduce → isolate → smallest fix → verify, instead of guessing a cause and patching code.
Use when the user asks to review a pull request, review a diff, review changes on a branch, or audit code written by a teammate or AI agent. Reviews the diff (not the summary) for unrelated changes, missing verification, behavior risks, and over-engineering.
Use when the user asks to add a new feature, add a new API/endpoint/page/component, implement a new capability, or extend existing functionality with new behavior. Forces a plan-before-code loop, scoping non-goals, and verifying the new behavior end-to-end.
Use when the user asks to understand, explain, trace, audit, or explore code WITHOUT modifying it. Examples — "how does X work", "where is Y called", "why does this branch exist", "trace the request flow". Stays read-only and does not produce a fix.
Use when a request is large, ambiguous, or spans multiple files/components and the user has NOT yet approved an implementation approach. Produces a verifiable plan via Plan Mode and a TodoWrite breakdown, then hands off to a more specific skill (bug-fix, refactor, feature-add) for execution.
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
The operational layer for coding agents. Bookkeeping, validation, and flows that compound knowledge between sessions.
An umbrella collection of 20 skills + 1 hook for AI-driven development: agent self-configuration (MCP / hooks / settings / subagents / skills / plugins / Claude Code optimisation), engineering practices (prompt engineering, FPF problem solving, bug-fix protocol, repository-history investigation, C# refactoring), research & docs (Semantic Scholar deep research, URL-to-Markdown, ubiquitous-language thesaurus), multi-agent orchestration (consilium), macOS health & cleanup, and niche utilities (clipboard, repo-explorer, Windows QA). Plus a Claude Code Bash safety hook.
Harness for Claude Code — skills, /harness:* slash commands, persona subagents, lifecycle hooks, and MCP tools without per-repo `harness setup`. Sibling plugins exist for Cursor, Gemini CLI, and Codex.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Iterative Ralph loops with Codex CLI review gates at completion. Claude works until done, then Codex reviews before allowing exit.
Verification-first engineering toolkit for Claude Code. 15 skills across a 5-phase spine (Investigate → Design → Implement → Verify → Ship), 8 specialist agents, an interactive setup wizard. Every skill has rationalizations + evidence requirements. Built for senior ICs and tech leads.
AI Coding Agent 越来越强。
但真正的问题不是它不会写代码。
真正的问题是:
这个项目想做一件事:
把工程师脑子里的工程判断,沉淀成 Agent 能执行的规则、Skill、检查清单和工作流。
这里不是 Prompt 收藏夹。
这里是 AI Coding 的工程实践手册。
让 AI 不只是会写代码,而是按一个靠谱工程师的方式写代码。
英文定位:
Turn engineering judgment into instructions that coding agents can follow.
很多人第一次使用 AI Coding Agent 时,都会经历一个很爽的阶段。
你说一句:
帮我把这个 bug 修一下。
Agent 开始读代码、改文件、生成 diff、写总结。
看起来很聪明。
但时间久了,你会发现几个非常真实的问题:
所以 AI Coding 的核心不是让模型多写代码。
更重要的是:
怎么让 Agent 不乱写代码。
agent-coding-playbook/
├── README.md
├── CLAUDE.md
├── AGENTS.md
├── LICENSE
├── docs/
│ ├── ai-coding-maturity-model.md
│ ├── context-engineering.md
│ ├── mcp-tooling-guidelines.md
│ ├── mcp-vs-skill.md
│ └── vibe-coding-risk-level.md
├── skills/
│ ├── acp-bug-fix/SKILL.md
│ ├── acp-code-review/SKILL.md
│ ├── acp-refactor/SKILL.md
│ ├── acp-release-check/SKILL.md
│ └── acp-test-first/SKILL.md
├── checklists/
│ ├── before-coding.md
│ ├── before-commit.md
│ ├── high-risk-change.md
│ └── pr-review.md
├── examples/
│ ├── bad-prompts.md
│ └── good-prompts.md
└── templates/
├── agents-basic.md
├── claude-basic.md
└── skill-template.md
当前版本聚焦六类资产:
CLAUDE.md:给 Claude Code 使用的项目级行为规则;AGENTS.md:给 Codex、Cursor、其他 coding agents 使用的通用规则;skills/:面向具体任务的 Agent 工作流;checklists/:给人类和 Agent 共用的检查清单;docs/:AI Coding 工程化方法说明;examples/:好坏任务描述示例。写代码之前,先想清楚。
Agent 需要先说明:
如果需求不清楚,就不要假装清楚。
优先选择最简单、最直接、最小可行的实现。
不要为了一个小问题引入:
能用 50 行解决的问题,不要写成 200 行。
像外科手术一样改代码。
只改必须改的地方。
每一行 diff 都应该能回答这个问题:
这一行为什么和当前需求有关?
如果回答不上来,就不应该改。
不要只告诉 Agent “做什么”。
要告诉它:
做到什么才算完成。
好的完成标准包括:
没有验证,不要宣布完成。
上下文不是越多越好。
Agent 不应该把整个仓库、所有文档、所有历史信息都塞进一次会话。
更好的做法是:
Agent 的总结不是事实。
真正的事实是 diff、测试结果和运行结果。
对于高风险改动,必须人工 Review。
高风险范围包括:
这个仓库本身是一个 Claude Code 插件 + marketplace。安装后,Skill 会按需自动触发,并附带 /acp-init 一键初始化项目规则。
在 Claude Code 里执行两步:
# 1. 把本仓库注册为 marketplace(marketplace 名字叫 acp)
/plugin marketplace add bravekingzhang/agent-coding-playbook
# 2. 从 acp marketplace 安装 agent-coding-playbook 插件
/plugin install agent-coding-playbook@acp
注意:Claude Code 没有
/plugin install user/repo这种一步式语法。必须先marketplace add,再install <plugin>@<marketplace>。
本地开发或离线安装:
git clone https://github.com/bravekingzhang/agent-coding-playbook.git ~/agent-coding-playbook
# 在 Claude Code 中
/plugin marketplace add ~/agent-coding-playbook
/plugin install agent-coding-playbook@acp
安装后可用:
acp- 命名空间避免冲突):acp-bug-fix, acp-code-review, acp-refactor, acp-release-check, acp-test-first, acp-feature-add, acp-investigate, acp-plan-task/acp-init —— 在当前项目根写入 CLAUDE.md 模板并尝试自动检测构建/测试命令把 CLAUDE.md 或 AGENTS.md 复制到你的项目根目录。
然后根据你的项目特点补充:
如果你正在修 bug,可以使用:
skills/acp-bug-fix/SKILL.md
如果你正在做 Code Review,可以使用:
skills/acp-code-review/SKILL.md
如果你正在发布版本,可以使用:
skills/acp-release-check/SKILL.md
如果你希望先写测试再实现,可以使用:
skills/acp-test-first/SKILL.md
例如 PR Review 时,可以使用:
checklists/pr-review.md
发布前,可以使用:
checklists/high-risk-change.md
这个项目适合:
它不是:
它更像一本工作手册:
教你把工程经验写成 Agent 能遵守的规则。
Prompt 是临场发挥,Playbook 才是团队资产。
或者更直接一点:
不要训练 AI 会写代码,要训练 AI 按工程规矩写代码。
The initial version includes: