mnzralee/claude-multi-agent-architecture
A production-ready, reusable template for building sophisticated multi-agent AI systems with Claude Code. This architecture enables hierarchical agent coordination, specialized task delegation, and comprehensive documentation workflows.
ARCHETYPE
Fort Knox
Deny lists, pre-tool hooks, sandboxing. Nothing touches prod without a signature.
GUARDRAILS
5/5
Blocks destructive commands · Protects secrets · Pre-tool screening hook · No YOLO mode · Sandbox or ask-first rules · details
Copy this rig
# review before running: this installs third-party code
$ npx degit mnzralee/claude-multi-agent-architecture/.claude ./rig-claude-multi-agent-architecture # 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.
$ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/api-design .claude/skills/api-design $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/backend .claude/skills/backend $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/caveman .claude/skills/caveman $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/commit .claude/skills/commit $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/evaluator-optimizer .claude/skills/evaluator-optimizer $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/frontend-design-system .claude/skills/frontend-design-system $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/interactive-live-testing .claude/skills/interactive-live-testing $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/module-planning .claude/skills/module-planning $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/multi-agent-orchestration .claude/skills/multi-agent-orchestration $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/plan-feature .claude/skills/plan-feature $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/pr .claude/skills/pr $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/ralph-loop .claude/skills/ralph-loop $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/review-board .claude/skills/review-board $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/review .claude/skills/review $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/systematic-debugging .claude/skills/systematic-debugging $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/tdd-workflow .claude/skills/tdd-workflow $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/verification-before-completion .claude/skills/verification-before-completion $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/verification .claude/skills/verification $ npx degit mnzralee/claude-multi-agent-architecture/.claude/skills/work-recording .claude/skills/work-recording
$ curl -fsSL --create-dirs -o .claude/agents/architect.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/architect.md $ curl -fsSL --create-dirs -o .claude/agents/backend-impl.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/backend-impl.md $ curl -fsSL --create-dirs -o .claude/agents/code-quality-auditor.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/code-quality-auditor.md $ curl -fsSL --create-dirs -o .claude/agents/cqrs-specialist.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/cqrs-specialist.md $ curl -fsSL --create-dirs -o .claude/agents/db-specialist.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/db-specialist.md $ curl -fsSL --create-dirs -o .claude/agents/debugger.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/debugger.md $ curl -fsSL --create-dirs -o .claude/agents/docker-deploy.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/docker-deploy.md $ curl -fsSL --create-dirs -o .claude/agents/e2e-tester.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/e2e-tester.md $ curl -fsSL --create-dirs -o .claude/agents/evaluator.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/evaluator.md $ curl -fsSL --create-dirs -o .claude/agents/frontend-impl.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/frontend-impl.md $ curl -fsSL --create-dirs -o .claude/agents/infra-impl.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/infra-impl.md $ curl -fsSL --create-dirs -o .claude/agents/prompt-writer.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/prompt-writer.md $ curl -fsSL --create-dirs -o .claude/agents/researcher.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/researcher.md $ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/security.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/security.md $ curl -fsSL --create-dirs -o .claude/agents/supervisor.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/supervisor.md $ curl -fsSL --create-dirs -o .claude/agents/tester.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/tester.md $ curl -fsSL --create-dirs -o .claude/agents/watchdog.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/watchdog.md $ curl -fsSL --create-dirs -o .claude/agents/work-recorder.md https://raw.githubusercontent.com/mnzralee/claude-multi-agent-architecture/main/.claude/agents/work-recorder.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": [
"Bash(rm -rf:*)",
"Bash(rm -r:*)",
"Bash(git push:*)",
"Bash(git reset --hard:*)",
"Bash(git checkout .:*)",
"Bash(git restore .:*)",
"Bash(git clean:*)",
"Bash(git rebase:*)",
"Bash(curl:*)",
"Bash(wget:*)",
"Read(.env)",
"Read(**/.env)",
"Read(**/.env.*)",
"Read(**/secrets/**)",
"Read(**/*secret*)",
"Read(**/*credential*)",
"Read(**/id_rsa)",
"Read(**/id_ed25519)",
"Read(**/*.pem)",
"Read(**/*.key)"
],
"ask": [
"Bash(git checkout:*)",
"Bash(git switch:*)"
]
},
"hooks": {
"PreToolUse": [
{
"matcher": "Write|Edit|NotebookEdit",
"hooks": [
{
"type": "command",
"command": "python \"$CLAUDE_PROJECT_DIR/.claude/hooks/file-guard.py\""
}
]
}
]
}
} Skills (19)
api-designbackendcavemancommitevaluator-optimizerfrontend-design-systeminteractive-live-testingmodule-planningmulti-agent-orchestrationplan-featureprralph-loopreview-boardreviewsystematic-debuggingtdd-workflowverification-before-completionverificationwork-recording
Subagents (19)
| architect model: opus | Use for system design, module planning, API contracts, and architectural decisions BEFORE implementation. Invoke proactively for complex multi-file features. Read-only: produces designs and plans, nev |
| backend-impl model: sonnet | Use to implement backend code following clean architecture (domain -> application -> infrastructure -> interface). Writes use cases, repositories, services, and their tests; proves work with verbatim |
| code-quality-auditor model: opus | Use for deep, plan-only quality analysis: complexity, coupling, dead code, and a weighted health score with a phased remediation plan. Read-only. |
| cqrs-specialist model: sonnet | Use for event-driven and CQRS work: command/outbox writes, idempotent projectors, and read-model handlers. Domain-neutral; never catches errors inside a transaction boundary. |
| db-specialist model: sonnet | Use for database schema, migrations, and ORM operations. Knows generate-vs-migrate semantics, data-integrity gotchas, and never pushes schema to production or casts the client to any. Keep ORM-neutral |
| debugger model: sonnet | Use to diagnose failures via root-cause analysis (5-Whys) before any fix. Distinguishes symptom from cause and returns the minimal reproducing evidence. |
| docker-deploy model: sonnet | Use to build and ship container images: multi-stage builds, transitive monorepo dependencies, digest pinning, and registry push. Verifies the running build matches the intended digest. |
| e2e-tester model: sonnet | Use for end-to-end and smoke testing of critical user paths across viewports. Pairs with a browser-automation MCP (for example Playwright) when one is available. |
| evaluator model: opus | Use as a read-only critic to score another agent's output against an explicit rubric before it is accepted. Pairs with the evaluator-optimizer loop and the verification-before-completion skill. Return |
| frontend-impl model: sonnet | Use to implement frontend/UI code: components, pages, data fetching, forms, and client state. Follows accessibility and type-safety standards. |
| infra-impl model: sonnet | Use for infrastructure and deployment work: containers, orchestration manifests, health probes, network policy, and secrets. Dry-run first; never mutate production without explicit approval. |
| prompt-writer model: haiku | Use before dispatching any non-trivial sub-agent. Generates a context-rich, bounded prompt: task scope, verified file paths, acceptance criteria as commands, forbidden actions, and the output contract |
| researcher model: haiku | Use proactively for fast codebase exploration and pattern discovery. Returns the top 3-5 relevant files or patterns with a short summary. Read-only and optimized for speed. |
| reviewer model: sonnet | Use proactively to review code for correctness, security, and quality against the kit rules. Read-only; reports findings by severity with file:line evidence. |
| security model: opus | Use proactively for security audits of auth, input handling, secrets, and sensitive flows. Read-only; reports vulnerabilities with OWASP mapping and severity. |
| supervisor model: sonnet | Use to coordinate multi-track work and dispatch sub-agents. Manages task breakdown, sequencing, and synthesis of sub-agent results. Calls the prompt-writer first for every non-trivial dispatch. |
| tester model: sonnet | Use to write and run unit and integration tests following the testing pyramid and AAA structure. Proves results with verbatim test output, never a narrative claim. |
| watchdog model: sonnet | Use as a parallel safety monitor during large refactors or multi-agent runs. Watches for security, breaking-change, dependency, and quality regressions and recommends pausing on critical drift. Read-o |
| work-recorder model: haiku | Use to maintain the session work record under docs/workrecords/. Appends a first-person narrative of what was done, decided, and verified. Background documentation role. |
Hooks (3)
| event | matcher | runs |
|---|---|---|
| PreToolUse | Write|Edit|NotebookEdit | python "$CLAUDE_PROJECT_DIR/.claude/hooks/file-guard.py" |
| PostToolUse | Write|Edit | python "$CLAUDE_PROJECT_DIR/.claude/hooks/auto-format.py" |
| Stop | * | python "$CLAUDE_PROJECT_DIR/.claude/hooks/verification-gate.py" |
Permissions
deny (20)
Bash(rm -rf:*)
Bash(rm -r:*)
Bash(git push:*)
Bash(git reset --hard:*)
Bash(git checkout .:*)
Bash(git restore .:*)
Bash(git clean:*)
Bash(git rebase:*)
Bash(curl:*)
Bash(wget:*)
Read(.env)
Read(**/.env)
Read(**/.env.*)
Read(**/secrets/**)
Read(**/*secret*)
Read(**/*credential*)
Read(**/id_rsa)
Read(**/id_ed25519)
Read(**/*.pem)
Read(**/*.key)
ask (2)
Bash(git checkout:*)
Bash(git switch:*)
allow (34)
Read
Glob
Grep
WebSearch
WebFetch(domain:code.claude.com)
WebFetch(domain:docs.claude.com)
WebFetch(domain:www.anthropic.com)
WebFetch(domain:github.com)
WebFetch(domain:stackoverflow.com)
Bash(npm run:*)
Bash(npm test:*)
Bash(npm run lint:*)
Bash(npm run build:*)
Bash(npx tsc:*)
Bash(npx vitest:*)
Bash(npx jest:*)
Bash(git status:*)
Bash(git diff:*)
Bash(git log:*)
Bash(git show:*)
Bash(git blame:*)
Bash(git branch:*)
Bash(git add:*)
Bash(git commit:*)
Bash(git fetch:*)
Bash(ls:*)
Bash(cat:*)
Bash(head:*)
Bash(tail:*)
Bash(wc:*)
Bash(tree:*)
Bash(find:*)
Bash(grep:*)
Bash(rg:*)
Similar rigs
ysoftman/myenv
personal environment settings
Skill Collector 2.0k tok ·
yourconscience/tackroom
Dotfiles for your AI agents: one ~/.agents repo for skills, MCP servers, hooks, agent roles, and memory — synced natively to Claude Code, Codex, OMP/Pi, OpenCode, Droid, and Hermes. Commit-pinned and audited externals.
Orchestrator 1.4k tok ·
justxor/claude-code-pro-course
Claude Code PRO — полный курс и гайд на русском: CLAUDE.md, settings, skills, subagents, hooks, MCP, плагины, CI. 8 лабораторных, библиотека промптов и готовый starter-kit. Здесь бесплатные курсы по chatgpt -
Fort Knox 27.4k tok ·
bar181/swarm-setup
Transform your AI development with custom agent swarms optimized for Agentic Engineering - optimal settings for Claude Code and Claude-flow
Orchestrator 13.7k tok ·