Dual Brain

What This Skill Does

The Dual Brain skill is a meta-cognitive enhancement tool for OpenClaw agents. By utilizing a secondary, independent LLM to process every incoming user message, the skill generates a brief, 2-3 sentence 'perspective' or critique. This perspective is stored locally and read by your primary agent before it generates a final response. By creating a heterogeneous feedback loop, you mitigate model-specific biases, gain alternative technical approaches, and increase the factual robustness of your agent's interactions. The background daemon handles all API calls, ensuring the primary agent experience remains fluid and responsive.

Installation

To integrate Dual Brain into your agent ecosystem, first ensure you have the necessary runtime environment for Node.js. Run the following command in your terminal: clawhub install openclaw/skills/skills/dannydvm/dual-brain. Once installed, use the CLI tool to initialize your preferences: dual-brain setup. This will trigger an interactive prompt to select your secondary model provider (Ollama, Moonshot, OpenAI, or Groq) and input your credentials. Finally, initiate the background process with dual-brain start or use dual-brain install-daemon to ensure the skill persists across system reboots.

Use Cases

Dual Brain is ideal for complex decision-making, creative brainstorming, and code review. It excels in scenarios where a 'single-model' bias could be detrimental, such as drafting technical documentation, debugging edge-case code, or debating philosophical topics. By forcing the inclusion of a second model's view, the agent becomes a collaborative pair-programmer rather than a static echo chamber.

Example Prompts

1. "Analyze this Python script for performance bottlenecks and provide a list of recommended refactoring steps."
2. "Draft a professional response to this aggressive client email while maintaining a tone of diplomatic conciliation."
3. "Explain the implications of the latest React updates, highlighting potential breaking changes for legacy codebases."

Tips & Limitations

To get the most out of Dual Brain, choose a secondary model that has different architectural origins than your primary. For instance, if your primary is GPT-4, using a local Llama-3 model via Ollama provides excellent cognitive contrast. Keep in mind that this skill introduces additional API costs if using commercial providers, so monitor your usage via the dual-brain logs command. If the secondary LLM is slow, it may increase your total latency by a few hundred milliseconds; if high-speed performance is required, prefer local models.