ask-council

What This Skill Does

The ask-council skill serves as a headless interface to the powerful LLM Council backend within the OpenClaw ecosystem. Rather than requiring users to switch contexts or open a dedicated web UI to evaluate complex problems, this skill enables a conversational workflow. It triggers a multi-stage process where multiple AI models provide independent responses, critique one another, and ultimately arrive at a consensus. The process is managed by the Chairman agent, which synthesizes the final output, providing a high-confidence, filtered answer directly inside your Telegram or chat interface.

Installation

To integrate this skill into your OpenClaw agent, ensure the LLM Council backend is pre-installed. Execute the following command in your terminal:

clawhub install openclaw/skills/skills/jeadland/ask-council

Before running the skill, verify that your backend service is active by running /install-llm-council. If the backend is unreachable, the skill is designed to prompt you to initialize the service, ensuring a smooth user experience.

Use Cases

This skill is perfect for scenarios requiring multi-perspective reasoning without the overhead of manual data synthesis. Ideal use cases include:

• Technical Decision Making: Deciding between architecture patterns, database technologies, or programming languages.
• Market Analysis: Evaluating broad business questions or investment queries where multiple model viewpoints mitigate individual model hallucination.
• Complex Problem Solving: Brainstorming solutions for multi-faceted problems that benefit from adversarial debate among different specialized AI models.
• On-the-go Research: Getting synthesized, high-quality advice while mobile, where accessing a full web-based dashboard is impractical.

Example Prompts

1. "/council Should I switch my backend infrastructure to Rust for better performance?"
2. "/council Is a flat organizational structure more effective than a hierarchical one for a startup?"
3. "/council What are the primary security implications of using public cloud versus on-premise storage?"

Tips & Limitations

• Patience is Key: Because the skill orchestrates a multi-step deliberation process (generation, critique, and synthesis), it typically requires 30-60 seconds to provide an answer. Do not interrupt the process once it begins.
• State Management: This skill creates a new conversation context for every request. It does not maintain long-term memory across multiple council calls.
• Deep Dives: If the Chairman's answer is insufficient or you want to see the specific reasoning used by individual models, always follow the link provided in the output to the full Web UI (typically hosted at http://10.0.1.184:5173).