smithdak/base
Define your agentic system once — rules, agents, pipelines, gates, knowledge — and compile it into native config for Claude Code, Codex, and GitHub Copilot. Plain files in git, ▎ a Rust CLI that never runs the agent loop, and durable work/
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
GUARDRAILS
3/5
Blocks destructive commands · Pre-tool screening hook · No YOLO mode · details
Quality 22/100 Guardrails v2
Copy this rig
# review before running: this installs third-party code $ claude mcp add --transport http github https://api.githubcopilot.com/mcp/ $ npx degit smithdak/base/.claude ./rig-base # 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 github https://api.githubcopilot.com/mcp/ $ npx degit smithdak/base/.agents/skills/automation .claude/skills/automation $ npx degit smithdak/base/.agents/skills/decision-record .claude/skills/decision-record $ npx degit smithdak/base/.agents/skills/delivery .claude/skills/delivery $ npx degit smithdak/base/.agents/skills/durable-handoff .claude/skills/durable-handoff $ npx degit smithdak/base/.agents/skills/evidence-review .claude/skills/evidence-review $ npx degit smithdak/base/.agents/skills/fix .claude/skills/fix $ npx degit smithdak/base/.agents/skills/migrate .claude/skills/migrate $ npx degit smithdak/base/.agents/skills/pickup .claude/skills/pickup $ npx degit smithdak/base/.agents/skills/research .claude/skills/research $ npx degit smithdak/base/.agents/skills/writing .claude/skills/writing
$ curl -fsSL --create-dirs -o .claude/agents/builder.md https://raw.githubusercontent.com/smithdak/base/main/.claude/agents/builder.md $ curl -fsSL --create-dirs -o .claude/agents/delivery-analyst.md https://raw.githubusercontent.com/smithdak/base/main/.claude/agents/delivery-analyst.md $ curl -fsSL --create-dirs -o .claude/agents/delivery-auditor.md https://raw.githubusercontent.com/smithdak/base/main/.claude/agents/delivery-auditor.md $ curl -fsSL --create-dirs -o .claude/agents/delivery-implementer.md https://raw.githubusercontent.com/smithdak/base/main/.claude/agents/delivery-implementer.md $ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.com/smithdak/base/main/.claude/agents/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": [
"Bash(git push * main*)",
"Bash(git push * HEAD:main*)"
]
},
"hooks": {
"PreToolUse": [
{
"matcher": "^(?:Bash)$",
"hooks": [
{
"type": "command",
"command": "base"
}
]
},
{
"matcher": "^(?:Edit|Write|NotebookEdit)$",
"hooks": [
{
"type": "command",
"command": "base"
}
]
},
{
"matcher": "^(?:mcp__github__.*)$",
"hooks": [
{
"type": "command",
"command": "base"
}
]
}
]
}
} MCP servers (1)
| server | source | est. tokens |
|---|---|---|
| GitHub MCP · "github" | remote · remote | 18.0k |
Skills (10)
Subagents (5)
| builder model: inherit | Implements approved plans and verifies the resulting behavior. |
| delivery-analyst model: inherit | Maps the requested outcome to repository reality before a plan is approved. |
| delivery-auditor model: inherit | Challenges a change against its acceptance checks, risks, and retained evidence. |
| delivery-implementer model: inherit | Implements an approved, bounded change and retains verification evidence. |
| reviewer model: inherit | Reviews changes for correctness, regressions, safety, and missing verification. |
Hooks (4)
| event | matcher | runs |
|---|---|---|
| PreToolUse | ^(?:Bash)$ | base |
| PreToolUse | ^(?:Edit|Write|NotebookEdit)$ | base |
| PreToolUse | ^(?:mcp__github__.*)$ | base |
| SessionStart | * | base |
Permissions
deny (2)
Bash(git push * main*)
Bash(git push * HEAD:main*)
ask (0)
—
allow (0)
—
Similar rigs
Gildaciolopes/agentic-dev-stack
Smart vibe coding with Claude Code: multi-agent orchestration (Opus plans; Sonnet, Haiku, Gemini, Codex and free models execute), Spec Kit SDD, caveman + RTK + graphify token savings, sandboxed secrets, self-healing free models and a Hermes
Minimalist 1.2k tok · · Q10
brackenjoe/agent-infrastructure
A guide for AI agent set up. A manager, a cheaper builder for code, any agents for other repetitive tasks, and the governance to keep them in line.
Minimalist 1.4k tok · · Q18
aniteshdharam/claude-token-policy
A Claude Code plugin that cuts your token bill by shrinking context per call, not by picking a cheaper model: a policy skill, enforcing hooks, and a census script that measures your own transcripts.
Minimalist 247 tok · · Q0
Freakling/App-Director
Director-oriented AI coding workflow for small apps. You direct the design, AI builds it. Works for web, mobile, and desktop apps.
Fort Knox 2.3k tok · · Q41