agentrigs
+ submit rig
~ / rigs / Haven16262 / dual-model-workflow

Haven16262/dual-model-workflow

双模型协作开发工作流：一个当全局者、一个当工作者（默认示例接 DeepSeek），用三个 Markdown 文件协调 | A dual-model AI coding workflow — Overseer + Worker (DeepSeek by default), coordinated via three Markdown files

↗ GitHub ★ 1 MIT updated 9d ago project Claude Code
share on X
ARCHETYPE
Pragmatist
A balanced, no-drama setup: some rules, some tools, nothing extreme.
CONTEXT TAX · EVERY TURN
~107 tokens
Featherweight · median rig: 2.3k · breakdown
GUARDRAILS
1/5
No YOLO mode · details

Copy this rig

# review before running: this installs third-party code
$ npx degit Haven16262/dual-model-workflow/templates/.claude ./rig-dual-model-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.

$ npx degit Haven16262/dual-model-workflow/skills/allow-once .claude/skills/allow-once
$ curl -fsSL --create-dirs -o .claude/agents/critic.md https://raw.githubusercontent.com/Haven16262/dual-model-workflow/main/templates/.claude/agents/critic.md

This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 7,204 rigs:

{
  "permissions": {
    "deny": [
      "Read(./.env)",
      "Read(**/.env)",
      "Read(~/.ssh/**)",
      "Bash(rm -rf *)",
      "Read(**/*.pem)",
      "Bash(rm -rf /)",
      "Bash(git push --force:*)",
      "Bash(sudo *)",
      "Read(.env)",
      "Bash(rm -rf /*)",
      "Read(./.env.*)",
      "Read(~/.aws/**)",
      "Bash(git push --force*)",
      "Bash(rm -rf:*)",
      "Read(**/*.key)",
      "Read(**/.env.*)",
      "Bash(sudo:*)",
      "Bash(git reset --hard*)",
      "Bash(git reset --hard:*)",
      "Read(.env.*)"
    ],
    "ask": [
      "Bash(git push:*)",
      "Bash(git push *)",
      "Bash(git commit:*)",
      "Bash(rm *)",
      "Bash(rm:*)",
      "Bash(git rebase *)",
      "Bash(wget *)",
      "Bash(npm publish:*)",
      "Bash(git commit *)",
      "Bash(gh pr merge *)"
    ]
  }
}

Skills (1)

allow-once

Subagents (1)

critic
model: haiku
双模型工作流的安全审查子代理。由全局者在工作者「关键决策点」的「安全相关」项非空时调用,对工作者点名的相关文件做结构化审查并返回报告。只读,不可修改任何文件。

Hooks (2)

eventmatcherruns
Stop*f="$HOME/.claude/scripts/k10-stop-gate.py"; if [ -f "$f" ]; then for py in python3 python; do if "$py" -c "" >/dev/null 2>&1; then exec "$py" "$f"; fi; done; printf '{"systemMessage":"K10 钩子找不到能运行的 Python(python3 / python 都跑不起来),本次不检查"}\n';
SubagentStop*f="$HOME/.claude/scripts/k10-subagent-report.py"; if [ -f "$f" ]; then for py in python3 python; do if "$py" -c "" >/dev/null 2>&1; then exec "$py" "$f"; fi; done; printf '{"systemMessage":"K10 钩子找不到能运行的 Python(python3 / python 都跑不起来),本次不检查

Slash commands (2)

/as-overseer/as-worker

Similar rigs

SchotjeChrisman/agentic-workflow
A self-improving Claude Code setup
★ 0 setup Claude Code 5 agents 6 hooks 1d ago
Orchestrator 2.8k tok ·
navox-labs/agents
15-agent AI engineering team for Claude Code. Full sprint cycle: Think, Plan, Build, Review, Test, Ship, Reflect. Zero dependencies.
★ 13 collection Claude Code 19 agents 4mo ago
Orchestrator 5.1k tok ·
TomSzenessy/AI_TEMPLATE
Self-healing starting point for AI-agent-built projects of any size: docs bound to code, generated host adapters, delegation roles, and deterministic gates for any agent or model.
★ 0 project Claude CodeCodexGemini CLI 3 mcp 11 skills 6 agents 4 hooks today
Orchestrator 12.7k tok ·
angadhn/botference
One room, three minds: a multi-LLM planning council for your terminal — you, Claude Code, and Codex debate and ship an implementation plan.
★ 19 project Claude Code 6 skills 19 agents 5d ago
Orchestrator 700 tok ·
copied ✓