agentskillexchange Escalate risky coding tasks into multi-model debate and consensus gates with Claude Octopus
Use Claude Octopus when ordinary Claude Code flows are not enough and a risky coding task needs adversarial multi-model review, disagreement surfacing, and a consensus gate before shipping.
Prerequisites
Claude Code; optional additional model providers such as Codex, Gemini, Copilot, Qwen, Ollama, Perplexity, or OpenRouter
Installation
Use the upstream install or setup path that matches your environment:
- npm install -g @openai/codex @google/gemini-cli @qwen-code/qwen-code 2>/dev/null || true
- npm install github:nyldn/claude-octopus#main --prefix openclaw
Requirements and caveats from upstream:
- Provider API calls require internet access from the hosted environment.
- "command": "node",
Basic usage or getting-started notes:
For scheduled Claude Code tasks, run /octo:sentinel for triage and /octo:security for recurring audits. Keep jobs read-only by default and route fixes through /octo:debug, /octo:review, or /octo:embrace after triage.
| Reduce token usage | /octo:doctor (includes RTK install + token tips) |
| Just run something quick | /octo:quick |
Extracted from upstream docs: https://raw.githubusercontent.com/nyldn/claude-octopus/HEAD/README.md