robertanton81/claude-subagent-router
Claude Code plugin with seven subagent workers, a review skill and a Codex CLI bridge. Optional model routing for Agent tool calls uses Jev, a paid classifier. Codex and Jev are off by default.
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 robertanton81/claude-subagent-router/agents ./rig-claude-subagent-router/agents $ npx degit robertanton81/claude-subagent-router/skills ./rig-claude-subagent-router/skills $ npx degit robertanton81/claude-subagent-router/hooks ./rig-claude-subagent-router/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 robertanton81/claude-subagent-router/skills/configure .claude/skills/configure $ npx degit robertanton81/claude-subagent-router/skills/delegate .claude/skills/delegate $ npx degit robertanton81/claude-subagent-router/skills/report .claude/skills/report $ npx degit robertanton81/claude-subagent-router/skills/review .claude/skills/review $ npx degit robertanton81/claude-subagent-router/skills/setup .claude/skills/setup
$ curl -fsSL --create-dirs -o .claude/agents/codex-implementer.md https://raw.githubusercontent.com/robertanton81/claude-subagent-router/main/agents/codex-implementer.md $ curl -fsSL --create-dirs -o .claude/agents/codex-reviewer.md https://raw.githubusercontent.com/robertanton81/claude-subagent-router/main/agents/codex-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/complete-searcher.md https://raw.githubusercontent.com/robertanton81/claude-subagent-router/main/agents/complete-searcher.md $ curl -fsSL --create-dirs -o .claude/agents/debugger.md https://raw.githubusercontent.com/robertanton81/claude-subagent-router/main/agents/debugger.md $ curl -fsSL --create-dirs -o .claude/agents/implementer.md https://raw.githubusercontent.com/robertanton81/claude-subagent-router/main/agents/implementer.md $ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.com/robertanton81/claude-subagent-router/main/agents/reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/searcher.md https://raw.githubusercontent.com/robertanton81/claude-subagent-router/main/agents/searcher.md
This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 16.2k rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(./.env.*)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(git push --force:*)",
"Read(**/*.pem)",
"Read(.env)",
"Read(~/.aws/**)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Bash(rm -rf *)",
"Bash(sudo *)",
"Bash(sudo:*)",
"Bash(rm -rf /)",
"Bash(git push -f:*)",
"Read(.env.*)",
"Bash(git reset --hard*)",
"Bash(rm -rf /*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(gh pr merge *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(gh pr merge:*)",
"Bash(gh pr create *)",
"Bash(git reset --hard:*)",
"Bash(npm publish:*)",
"Bash(git rebase *)"
]
}
} Skills (5)
Subagents (7)
| codex-implementer model: haiku | Runs an implementation task in the Codex CLI on the ChatGPT plan, so the work does not use Claude limits. Codex starts with no context. Use only with a complete brief that names the goal, the files, t |
| codex-reviewer model: haiku | Runs a Codex code review on the ChatGPT plan. By default it reviews the uncommitted changes. A line "review-scope:" in the brief picks another scope - base:<branch>, commit:<hash> or custom. Use after |
| complete-searcher model: sonnet | Finds and lists existing code or files when the answer is right only if the list names every match, such as "which directories exist under X" or "which files call Y". It changes no files. Use instead |
| debugger model: opus | Finds the cause of a failure, an error or wrong behaviour when the cause is not known yet, and fixes it when the brief asks for the fix. Use for failing tests, stack traces and regressions. |
| implementer model: sonnet | Writes and changes code inside a scope that the brief defines, from exact mechanical edits to features that need design choices. Use when files must change and the cause or the goal is already known. |
| reviewer model: sonnet | Reviews a set of code changes and reports problems with a priority, a file and a line. Use after a logical piece of work, above all when Codex wrote the change. It changes no files. |
| searcher model: haiku | Finds, reads and explains existing code or documentation, and reports what it found. Use for questions such as "where is X handled", "how does Y work" or "which files use Z". It changes no files. |
Similar rigs
SavenkoSy/sin-flow
Ticket-to-PR flow for a live codebase, as a Claude Code plugin: setup, sync, ingest, plan, exec, review and learning stages.
Orchestrator 1.4k tok · · Q3
orynvn/claude-workspace-setup
A project template for Claude Code CLI — opinionated, token-efficient, and built for multi-agent workflows.
Orchestrator 22.3k tok · · Q18
Saba-Var/claude-ward
A read-only tripwire that watches Claude Code's local config and reports suspicious changes.
Pragmatist 1.8k tok · · Q21
yopazerbot/brieltje
Smetlede and Briel (Lede, Belgium) in 3D: a walkable village in the browser, alone or with up to 16 friends. Explore, Paintball, Action, and endless waves of Zombies. Three.js and Blender, made with Claude Code and OpenAI Codex..
Pragmatist 4.7k tok · · Q8