ask-council
von jeadland
Web & Frontend DevelopmentKurzbeschreibung
Ask LLM Council a question directly from Telegram/chat — get the chairman's synthesized answer without opening.
Beschreibung
Was dieser Skill macht
Sends a question to a running LLM Council backend and returns the Chairman's synthesized answer via chat command. Runs through a three-stage deliberation process: individual model responses, peer ranking, and final synthesis. Takes 30-60 seconds per query.
Gives access to multi-model consensus from any chat interface without spinning up or navigating a web UI.
Wann du ihn einsetzt
- Quick architectural decision from mobile
- Getting a consensus answer without opening a browser
- Asking a tech or finance question mid-conversation
- Comparing language/framework tradeoffs on the fly
- Running a fast headless council query from Telegram
Beispiel-Workflow
User asks: /council Is Python or Go better for a new microservice?
- 1
Runs ask-council.sh with the question as argument
- 2
Creates a new conversation via the LLM Council API
- 3
Polls Stage 1 until all models have responded
- 4
Polls Stage 2 until models have ranked each other
- 5
Extracts and returns the Chairman's Stage 3 synthesis
Chairman's synthesized answer recommending Python for team velocity or Go for high-throughput services, plus a link to the full web discussion
Voraussetzungen
- LLM Council backend running locally (install via /install-llm-council)
Installation
clawhub install jeadland/ask-councilFühre den CLI-Befehl aus oder füge den Prompt in Claude Code, Cursor, Codex oder deinen Coding-Agenten ein.