qshanx/docs-governance
Claude Code、Codex 与 ChatGPT 的文档驱动开发治理插件：防文档腐烂、维护上下文与决策、契约协作、测试与回归证据。
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
Copy this rig
# review before running: this installs third-party code $ npx degit qshanx/docs-governance/agents ./rig-docs-governance/agents $ npx degit qshanx/docs-governance/commands ./rig-docs-governance/commands $ npx degit qshanx/docs-governance/skills ./rig-docs-governance/skills $ npx degit qshanx/docs-governance/hooks ./rig-docs-governance/hooks
MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ npx degit qshanx/docs-governance/skills/agent-entrypoints .claude/skills/agent-entrypoints $ npx degit qshanx/docs-governance/skills/change-impact .claude/skills/change-impact $ npx degit qshanx/docs-governance/skills/context-and-decisions .claude/skills/context-and-decisions $ npx degit qshanx/docs-governance/skills/contract-first .claude/skills/contract-first $ npx degit qshanx/docs-governance/skills/docs-governance .claude/skills/docs-governance $ npx degit qshanx/docs-governance/skills/living-docs-governance .claude/skills/living-docs-governance $ npx degit qshanx/docs-governance/skills/loop-design-check .claude/skills/loop-design-check $ npx degit qshanx/docs-governance/skills/module-regression .claude/skills/module-regression $ npx degit qshanx/docs-governance/skills/product-evolution .claude/skills/product-evolution $ npx degit qshanx/docs-governance/skills/test-collaboration .claude/skills/test-collaboration
$ curl -fsSL --create-dirs -o .claude/agents/backend-dev.md https://raw.githubusercontent.com/qshanx/docs-governance/main/agents/backend-dev.md $ curl -fsSL --create-dirs -o .claude/agents/contract-director.md https://raw.githubusercontent.com/qshanx/docs-governance/main/agents/contract-director.md $ curl -fsSL --create-dirs -o .claude/agents/docs-auditor.md https://raw.githubusercontent.com/qshanx/docs-governance/main/agents/docs-auditor.md $ curl -fsSL --create-dirs -o .claude/agents/docs-governor.md https://raw.githubusercontent.com/qshanx/docs-governance/main/agents/docs-governor.md $ curl -fsSL --create-dirs -o .claude/agents/frontend-dev.md https://raw.githubusercontent.com/qshanx/docs-governance/main/agents/frontend-dev.md $ curl -fsSL --create-dirs -o .claude/agents/product-docs-manager.md https://raw.githubusercontent.com/qshanx/docs-governance/main/agents/product-docs-manager.md $ curl -fsSL --create-dirs -o .claude/agents/regression-auditor.md https://raw.githubusercontent.com/qshanx/docs-governance/main/agents/regression-auditor.md
This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 7,204 rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(rm -rf *)",
"Read(**/*.pem)",
"Bash(rm -rf /)",
"Bash(git push --force:*)",
"Bash(sudo *)",
"Read(.env)",
"Bash(rm -rf /*)",
"Read(./.env.*)",
"Read(~/.aws/**)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(sudo:*)",
"Bash(git reset --hard*)",
"Bash(git reset --hard:*)",
"Read(.env.*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(rm:*)",
"Bash(git rebase *)",
"Bash(wget *)",
"Bash(npm publish:*)",
"Bash(git commit *)",
"Bash(gh pr merge *)"
]
}
} Skills (10)
agent-entrypointschange-impactcontext-and-decisionscontract-firstdocs-governanceliving-docs-governanceloop-design-checkmodule-regressionproduct-evolutiontest-collaboration
Subagents (7)
| backend-dev model: sonnet | 后端开发工人。只在后端目录里开发，接口实现一律以 CONTRACT.md 指向的机器契约为准（只读，不许改），并为接口写"返回必须符合契约"的校验测试（provider verification）。可在单会话里被 contract-director 派活，也可在自己的终端独立跑、只通过契约文件异步协作。想改接口形状时提"契约变更请求"给契约拥有者，绝不偷偷改实现。 |
| contract-director model: sonnet | 契约拥有者 / 集成对账者。唯一拥有并维护 CONTRACT.md（端与端的数据接口契约），先定契约，最后做集成对账。支持单会话多 agent（派活）和多终端各自跑（契约当异步媒介、不在线派活）两种模式。当项目分前端/后端或多个服务、需要各端照同一份接口开发、防字段漂移导致集成炸掉时使用。它不写业务代码，只守契约、（必要时）派活、对账。 |
| docs-auditor model: sonnet | Use this agent when the user asks to audit documentation governance, check whether CLAUDE.md / CLAUDE_MAP.md / PROJECT_STATUS.md / PROJECT_LOG.md drifted from the real project, verify governance befor |
| docs-governor model: sonnet | 项目文档 setup 编排与活文档治理执行者。首次配置新项目、讨论后项目或已有项目时，按共享 Skill 协调产品文档与 Agent 入口；维护期按需增量更新治理载体，不固定铺满四件套。 |
| frontend-dev model: sonnet | 前端开发工人。只在前端目录里开发，数据接口一律以 CONTRACT.md 指向的机器契约为准（只读，不许改）。可在单会话里被 contract-director 派活，也可在自己的终端独立跑、只通过契约文件跟其他端异步协作。需要契约里没有的字段时，提"契约变更请求"给契约拥有者，绝不自己假设接口形状。 |
| product-docs-manager model: sonnet | 产品文档管理员。用户要求登记原始材料、整理产品认知与规划、维护需求池与单需求、关联版本及验收复盘，或统一需求编号、来源和生效基线时使用。复用已有产品资料和任务入口，不代替产品负责人批准需求。 |
| regression-auditor model: sonnet | 模块回归审计员。读 REGRESSION.md 回归台账，定位本次改动涉及的模块，跑"本模块 + 全部下游"的回归验收命令，以退出码为终审，输出红绿审计摘要。只跑只报不修代码。在大项目改完一个模块、需要确认没牵连其他模块时使用（/regression-audit 触发）。 |
Slash commands (8)
/contract/governance-audit/governance-init/governance-retro/governance-setup/governance-sync/governance/regression-audit
Similar rigs
GinVlad/claude-saas-template
Claude Code template for building SaaS with agentic AI coordination. 5-stage workflow
Orchestrator 612 tok ·
Rohankundra2009/claude-agents-forge
Build and run production-ready Claude Code agents with full agent specs, including turn limits, permissions, memory, and tool orchestration.
Orchestrator 1.3k tok ·
runnithan/claudex-setup
Portable Claude Code and Codex configuration: slash commands, subagents, hooks, skills, and a transcript-to-lessons pipeline. Installable as a plugin.
Fort Knox 3.2k tok ·
Mationetap/claude-agents-forge
Production-hardened agent system for Claude Code — uses the full agent spec (maxTurns, permissionMode, disallowedTools, Task() orchestration, memory scopes, MCP integration)
Orchestrator 1.3k tok ·