🐙 Claude Octopus

Every AI model has blind spots. Claude Octopus supports ten external provider integrations — Codex, Gemini, Antigravity CLI, Copilot, Qwen, Ollama, Perplexity, OpenRouter, OpenCode, and Grok — alongside the built-in Claude Code host, with consensus gates that flag disagreements before you ship.

Claude-native first, Octopus for escalation. Use Claude-native /init, /review, and /security-review when Claude is enough. Use Octopus when you want multiple model opinions, adversarial review, or stricter multi-LLM workflows.

🐙 Research, build, review, and ship — with ten external providers checking the host's work. Say what you need, and the right workflow runs. Claude-native handles the ordinary path; Octopus handles the escalated path. A 75% consensus gate catches disagreements before they reach production. No single model's blind spots slip through.

🧠 Remembers across sessions. Integrates with claude-mem and agentmemory for persistent memory — past decisions, research, and context survive session boundaries.

⚡ Spec in, software out. Dark Factory mode takes a spec and autonomously runs the full pipeline — research, define, develop, deliver. You review the output, not every step.

🔄 Four-phase methodology, not just tools. Every task moves through Discover → Define → Develop → Deliver, with quality gates between phases. Other orchestrators give you infrastructure. Octopus gives you the workflows.

🐙 32 specialized personas (role-specific AI agents like security-auditor, backend-architect), 50 commands (slash commands you type), 58 skills (reusable workflow modules). Say "audit my API" and the right expert activates. Don't know the command? The smart router figures it out.

🐙 Works with just Claude. Adds up to ten external provider integrations. Zero external providers are needed to start. Add them one at a time — each activates automatically when detected.

💰 Five providers cost nothing extra when you already have access. Codex, Gemini, Antigravity CLI, and Copilot use existing subscriptions or local auth. Ollama runs locally for free. Qwen now requires API-key or Coding-Plan auth; its free OAuth tier ended on 2026-04-15.

What's New

🆕 v9.50 — Claude Code 2026 compatibility layer. Routine manifest for scheduled and GitHub-event automations, a SubagentStop quality/cost gate, /octo:usage per-provider cost attribution in Claude Code's /usage schema, a worktree.bgIsolation opt-out for fast direct-edit runs, a Claude Agent SDK seat (CLAUDE_SDK_API_KEY, Opus 4.8, 1M context), a four-skill starter pack, and a /plugin browse manifest with projected context cost.

/octo:usage --format table        # who spent what, per provider/skill/MCP server
OCTOPUS_WORKTREE_BG_ISOLATION=false   # skip worktree cloning for fast runs

🆕 v9.41 — Multi-LLM Council. /octo:council runs a structured 3/5/7-persona deliberation across Claude, Codex, Gemini, and OpenCode with goal modes (advice, decision, plan, implement, review), styles (balanced, adversarial, red-team, executive, implementation), benchmark-aware role routing, quorum + critical-veto gates, budget caps, and gated worktree handoff for approved plans. Use it when one model's opinion isn't enough.

/octo:council --goal decision --style adversarial "Should this service stay monolithic?"
/octo:council --goal implement --implement plan-only "Refactor the auth flow"

Full changelog →

• The Codex seat's premium default moved from GPT-5.4 to GPT-5.5 (v9.44+). Pin the old model with OCTOPUS_CODEX_MODEL=gpt-5.4 if needed.
• New claude-sdk seat env vars (v9.50): CLAUDE_SDK_API_KEY, OCTOPUS_CLAUDE_SDK_MODEL, OCTOPUS_CLAUDE_SDK_MAX_TOKENS, OCTOPUS_CLAUDE_SDK_ALLOWED_MODELS, OCTOPUS_CLAUDE_SDK_CONTEXT_BUDGET.
• New Fable 5 guard env vars (v9.51): OCTOPUS_FABLE5_MODE (auto/off/on), OCTOPUS_FABLE5_NO_RETRY. Guards auto-enable only when you pin claude-fable-5.
• Premium Claude role routing (architect, strategist, security-reviewer to Opus) landed in v9.29; restore the older mapping with OCTOPUS_LEGACY_ROLES=1.

Quickstart

# Terminal (not inside a Claude Code session):
claude plugin marketplace add https://github.com/nyldn/plugins.git
claude plugin install octo@nyldn-plugins

# Then inside Claude Code:
/octo:setup

That's it. Setup detects installed providers, shows what's missing, and walks you through configuration. You need zero external providers to start — Claude is built in.

Claude Code v2.1.50+ is the minimum supported runtime. Newer Claude Code releases unlock additional Octopus diagnostics and release checks automatically; the current plugin tracks feature flags through Claude Code v2.1.157.

git clone --depth 1 https://github.com/nyldn/claude-octopus.git ~/.codex/claude-octopus && mkdir -p ~/.agents/skills && ln -sf ~/.codex/claude-octopus/skills ~/.agents/skills/claude-octopus

Restart Codex. Skills appear automatically — invoke with $skill-doctor, $skill-debug, etc.

Cursor uses Octopus as an MCP server (not a plugin — Cursor doesn't have Claude Code's plugin system). You get MCP tools like octopus_discover, octopus_review, etc. instead of /octo:* slash commands.

Important: Just cloning the repo is not enough. You must complete all three steps below — install dependencies and configure the MCP server — for Cursor to pick up Octopus tools.

# 1. Clone the repo
git clone --depth 1 https://github.com/nyldn/claude-octopus.git ~/.cursor/claude-octopus

# 2. Install MCP server dependencies
cd ~/.cursor/claude-octopus/mcp-server && npm install

# 3. Configure Cursor — add to ~/.cursor/mcp.json (global) or .cursor/mcp.json (per-project):

{
  "mcpServers": {
    "claude-octopus": {
      "command": "npx",
      "args": ["tsx", "${userHome}/.cursor/claude-octopus/mcp-server/src/index.ts"],
      "env": {
        "OCTO_CLAW_ENABLED": "true",
        "OPENAI_API_KEY": "${env:OPENAI_API_KEY}",
        "GEMINI_API_KEY": "${env:GEMINI_API_KEY}"
      }
    }
  }
}

Restart Cursor. Tools appear in Cursor's AI chat — invoke by asking e.g. "use octopus_discover to research X".

If you're running Cursor on Windows with WSL, clone the rep