micheleromano/agent-scaffold
Answer a few questions, get a working Claude Code setup: CLAUDE.md, settings, MCP, path-scoped rules, skills, subagents and hooks. Any stack.
stack: Python
ARCHETYPE
Fort Knox
Deny lists, pre-tool hooks, sandboxing. Nothing touches prod without a signature.
GUARDRAILS
4/5
Blocks destructive commands · Protects secrets · Pre-tool screening hook · No YOLO mode · details
Quality 57/100 Guardrails v2
Copy this rig
# review before running: this installs third-party code
$ npx degit micheleromano/agent-scaffold/.claude ./rig-agent-scaffold # 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 micheleromano/agent-scaffold/.claude/skills/ci .claude/skills/ci $ npx degit micheleromano/agent-scaffold/.claude/skills/plan .claude/skills/plan $ npx degit micheleromano/agent-scaffold/.claude/skills/review .claude/skills/review $ npx degit micheleromano/agent-scaffold/.claude/skills/ship .claude/skills/ship
$ curl -fsSL --create-dirs -o .claude/agents/code-reviewer.md https://raw.githubusercontent.com/micheleromano/agent-scaffold/main/.claude/agents/code-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/humane-design-reviewer.md https://raw.githubusercontent.com/micheleromano/agent-scaffold/main/.claude/agents/humane-design-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/test-writer.md https://raw.githubusercontent.com/micheleromano/agent-scaffold/main/.claude/agents/test-writer.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.*)"
]
},
"hooks": {
"PreToolUse": [
{
"matcher": "Bash",
"hooks": [
{
"type": "command",
"command": "python3 \"${CLAUDE_PROJECT_DIR}/.claude/hooks/block_dangerous.py\""
}
]
}
]
}
} Skills (4)
Subagents (3)
| code-reviewer model: sonnet | Reviews changes in agent-scaffold for correctness, readability and adherence to project conventions. Use after a feature or fix is written. |
| humane-design-reviewer model: sonnet | Reviews user-facing changes in agent-scaffold against the Center for Humane Technology\u0027s principles: dark patterns, attention and notification design, data and consent, accessibility, and AI tran |
| test-writer model: sonnet | Writes and extends tests for agent-scaffold. Use when a change needs test coverage. |
Hooks (2)
| event | matcher | runs |
|---|---|---|
| PreToolUse | Bash | python3 "${CLAUDE_PROJECT_DIR}/.claude/hooks/block_dangerous.py" |
| PostToolUse | Edit|Write | python3 "${CLAUDE_PROJECT_DIR}/.claude/hooks/format_on_edit.py" |
Permissions
deny (2)
Read(./.env)
Read(./.env.*)
ask (0)
—
allow (9)
Bash(uv run pytest)
Bash(uv run pytest *)
Bash(uv run ruff check .)
Bash(uv run ruff check . *)
Bash(uv run mypy)
Bash(uv run mypy *)
Bash(uv run ruff format *)
Bash(uv run agent-scaffold)
Bash(uv run agent-scaffold *)
Similar rigs
eiler2005/ai-tab-optimizer
Turns 100+ browser tabs into searchable knowledge: discriminated-union protocol (45 types), per-batch LLM failover, stop-and-resume across 1000+ tabs
Pragmatist 15.4k tok · · Q15
ByteBoyy/agentic-dev-orchestrator
Repository-native workflows for planning, delegating, reviewing, and shipping software with coding agents
Orchestrator 1.3k tok · · Q3
ribeirogab/specwright
A multi-host plugin for Claude Code, Codex, and OpenCode. It gives a repository one explicit, change-driven engineering workflow:
Pragmatist 1.6k tok · · Q0
nicodiansk/turbocharge
Opinionated engineering pipeline for Claude Code: 10 skills, 6 agents, 2 hooks, 1 workflow. Replaces ad-hoc agent sprawl with brainstorm → story → plan → build → review → ship.
Pragmatist 5.4k tok · · Q5