Standardizes development and deployment cycles through rule-based command orchestration and integrated logging.

Install

mkdir -p .claude/skills/project-os && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/4396" && unzip -o skill.zip -d .claude/skills/project-os && rm skill.zip

Installs to .claude/skills/project-os

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

AI project OS for autonomous loop, automated orchestration, and rule-driven execution.
86 charsno explicit “when” trigger
Intermediate

Key capabilities

  • Automate development processes
  • Orchestrate commands and workflows
  • Enforce rule-driven execution
  • Manage release workflows
  • Standardize verification and smoke testing
  • Track iterations and logs

How it works

This skill automates and orchestrates development processes by defining steps, commands, and rules. It ensures consistent execution and provides a traceable evidence chain for each stage.

Inputs & outputs

You give it
development process steps, rules, and commands
You get back
standardized development-to-deployment workflow with traceable evidence

When to use project-os

  • Standardize delivery processes across projects
  • Automate deployment and smoke testing
  • Manage release workflow documentation

About this skill

Project OS

用于在新项目中快速落地“开发规范 + 迭代日志 + 发布闭环”的通用体系(Project OS),面向未来演进为可自治、可编排、自动驱动研发流程的 AI 操作系统。

三段式定位

  1. 自治闭环
    覆盖需求—实现—验证—发布—线上冒烟—复盘的端到端闭环,系统自动推动流程完成并形成可追踪证据链。

  2. 自动化编排
    将研发流程拆解为可编排的步骤与指令(commands/skills/workflows),以最小人工干预串联执行、回滚与验收。

  3. 规则驱动执行
    所有行为由 Rulebook 统一约束与裁决,确保执行一致性、合规性与可审计性;如需例外必须显式声明并记录。

适用场景

  • 想把一套严格交付流程迁移到其他项目。
  • 需要在团队内统一验证/冒烟/发布规范。

快速落地

将本 Skill 的模板复制到目标项目根目录(若已有文件,需合并而非覆盖):

cp -R <skill>/assets/commands ./commands
cp -R <skill>/assets/docs ./docs
cp <skill>/assets/AGENTS.template.md ./AGENTS.md

关键约束

  • “完成所有/完成全部”默认执行完整上线闭环(migrations -> deploy -> 线上冒烟)。
  • 冒烟测试默认使用非仓库目录环境,禁止写入仓库子目录。
  • 每次开发阶段结束必须完成 build/lint/tsc + 冒烟(如适用)。

扩展与维护

  • 新增/修改指令:更新 commands/commands.md 并同步 AGENTS 索引。
  • 新增/修改规则:只在 AGENTS 的 Rulebook 维护。
  • 发布流程统一写入 docs/workflows/npm-release-process.md

模板索引

  • assets/AGENTS.template.md
  • assets/commands/commands.md
  • assets/docs/logs/TEMPLATE.md
  • assets/docs/logs/README.md
  • assets/docs/workflows/npm-release-process.md

Limitations

  • Smoke testing defaults to a non-repository directory environment
  • Smoke testing prohibits writing to repository subdirectories
  • Development phases must conclude with build/lint/tsc and smoke testing

How it compares

This skill provides a standardized, rule-driven, and automated framework for development processes, unlike manual or ad-hoc approaches.

Compared to similar skills

project-os side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
project-os (this skill)16moReviewIntermediate
agent-workflow-automation46moReviewAdvanced
Release03moReviewIntermediate
agentnote-release03moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry