zhuyihenzheng/claude-code-workflow
Claude Code 开发 AI 工作流搭建规划与实践手册(结合 2026 年社区最新实践)
ARCHETYPE
Fort Knox
Deny lists, pre-tool hooks, sandboxing. Nothing touches prod without a signature.
CONTEXT TAX · EVERY TURN
up to ~10.1k tokens
Claude Code loads MCP tools on demand, so the real cost is usually lower.
Heavy · median rig: 2.3k · breakdown
GUARDRAILS
4/5
Blocks destructive commands · Protects secrets · No YOLO mode · Sandbox or ask-first rules · details
Quality 26/100 Guardrails v2
Copy this rig
# review before running: this installs third-party code $ claude mcp add --transport http context7 https://mcp.context7.com/mcp $ claude mcp add playwright -- npx -y @playwright/mcp@0.0.85 --isolated $ npx degit zhuyihenzheng/claude-code-workflow/templates/project/.claude ./rig-claude-code-workflow # inspect, then merge into .claude/
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.
$ claude mcp add --transport http context7 https://mcp.context7.com/mcp $ claude mcp add playwright -- npx -y @playwright/mcp@0.0.85 --isolated
$ npx degit zhuyihenzheng/claude-code-workflow/.claude/skills/workflow-retro .claude/skills/workflow-retro $ npx degit zhuyihenzheng/claude-code-workflow/templates/project/.claude/skills/change-flow .claude/skills/change-flow $ npx degit zhuyihenzheng/claude-code-workflow/templates/project/.claude/skills/spring-endpoint .claude/skills/spring-endpoint $ npx degit zhuyihenzheng/claude-code-workflow/templates/user/skills/diagnosing-bugs .claude/skills/diagnosing-bugs $ npx degit zhuyihenzheng/claude-code-workflow/templates/user/skills/grilling .claude/skills/grilling $ npx degit zhuyihenzheng/claude-code-workflow/templates/user/skills/writing-for-agents .claude/skills/writing-for-agents
$ curl -fsSL --create-dirs -o .claude/agents/code-reviewer.md https://raw.githubusercontent.com/zhuyihenzheng/claude-code-workflow/main/templates/project/.claude/agents/code-reviewer.md Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"permissions": {
"deny": [
"Read(/**/.env)",
"Read(/**/.env.*)",
"Read(/**/*credentials*)",
"Read(/**/*secret*)",
"Edit(/**/.env)",
"Edit(/**/.env.*)",
"Edit(/**/*credentials*)",
"Edit(/**/*secret*)",
"Read(/**/*.xlsx)",
"Read(/**/*.xlsm)",
"Bash(rm -rf /*)",
"Bash(git push --force *)",
"Bash(git push -f *)",
"Bash(git push * --force)",
"Bash(git push * --force *)",
"Bash(git push * -f)",
"Bash(git push * -f *)",
"Read(/screen-tests/report.html)",
"Bash(node *run.mjs *--update-baseline*)",
"Bash(node *run.mjs *--update-visual*)"
],
"ask": [
"Edit(/src/main/resources/db/migration/**)",
"Edit(/screen-tests/baseline.json)",
"Bash(node *run.mjs *--engine ai*)",
"Bash(node *run.mjs *--engine midscene*)",
"Bash(node *run.mjs *--heal*)",
"Bash(node *run.mjs *--ai-assert*)"
]
}
} MCP servers (2)
| server | source | est. tokens |
|---|---|---|
| Context7 | remote · remote | 1.2k |
| Playwright MCP · "playwright" | npm | 7.5k |
Skills (6)
Subagents (1)
| code-reviewer model: sonnet | 独立したコードレビュー担当。機能実装の完了後、PR の作成前に呼び出す。SPEC/受入基準とプロジェクト規約に照らして diff を確認し、問題の報告のみを行い、コードは変更しない。 |
Permissions
deny (20)
Read(/**/.env)
Read(/**/.env.*)
Read(/**/*credentials*)
Read(/**/*secret*)
Edit(/**/.env)
Edit(/**/.env.*)
Edit(/**/*credentials*)
Edit(/**/*secret*)
Read(/**/*.xlsx)
Read(/**/*.xlsm)
Bash(rm -rf /*)
Bash(git push --force *)
Bash(git push -f *)
Bash(git push * --force)
Bash(git push * --force *)
Bash(git push * -f)
Bash(git push * -f *)
Read(/screen-tests/report.html)
Bash(node *run.mjs *--update-baseline*)
Bash(node *run.mjs *--update-visual*)
ask (6)
Edit(/src/main/resources/db/migration/**)
Edit(/screen-tests/baseline.json)
Bash(node *run.mjs *--engine ai*)
Bash(node *run.mjs *--engine midscene*)
Bash(node *run.mjs *--heal*)
Bash(node *run.mjs *--ai-assert*)
allow (24)
Bash(./mvnw -q compile)
Bash(./mvnw -B -q compile)
Bash(./mvnw test)
Bash(./mvnw verify)
Bash(mvnw.cmd -q compile)
Bash(mvnw.cmd -B -q compile)
Bash(mvnw.cmd test)
Bash(mvnw.cmd verify)
Bash(./mvnw.cmd -q compile)
Bash(./mvnw.cmd -B -q compile)
Bash(./mvnw.cmd test)
Bash(./mvnw.cmd verify)
PowerShell(.\mvnw.cmd -q compile)
PowerShell(.\mvnw.cmd -B -q compile)
PowerShell(.\mvnw.cmd test)
PowerShell(.\mvnw.cmd verify)
Bash(node run.mjs --engine playwright)
Bash(node run.mjs --engine playwright --build-only)
Bash(node run.mjs --engine playwright --only *)
Bash(git status)
Bash(git diff *)
Bash(git log *)
Bash(gh pr view *)
Bash(gh issue view *)
Similar rigs
PageAI-Pro/vibe-coding-starter
Create great UIs with AI, see the full tutorial
Pragmatist 12.5k tok · · Q0
AizenvoltPrime/claude-setup
—
Pragmatist 1.2k tok · · Q31
croffasia/cc-blueprint-toolkit
🏗️ Claude Code Plugin for smart blueprint-driven development
Pragmatist 8.7k tok · · Q0
codeverbojan/claude-code-kickstart
Production-grade agentic workflow template for Claude Code. 4-layer architecture, task playbooks, session memory, sub-agents, one-click install.
Orchestrator 10.2k tok · · Q15