| name | council-execution |
| description | Executes council queries by running the query pipeline across selected AI providers (Gemini, OpenAI, Grok, Perplexity), displaying formatted responses verbatim, and generating a synthesis of consensus, divergence, and recommendations. Invoked by the ask command during standard (non-agent) council queries. |
Council Query Execution
Step 1: Run Query and Save to File
bash ${CLAUDE_PLUGIN_ROOT}/scripts/run-council.sh --providers=gemini,openai -- "Your question"
This outputs the path to the saved file (e.g., .claude/council-cache/council-1734567890.md).
Flag syntax: Use = with no spaces: --providers=gemini,openai
CRITICAL: Always place -- before the prompt to prevent prompt text containing dashes from being parsed as flags.
Step 2: Read and Display the Output VERBATIM
Use the Read tool to read the output file path returned by Step 1.
CRITICAL: Display the file content EXACTLY as written. Do NOT:
- Reformat or reinterpret any text
- Add your own headers or structure
- Summarize or abbreviate responses
- Skip any lines including separator lines (
---)
Simply copy-paste the entire file content into your response.
Step 3: Complete the Synthesis Section
The file ends with a ## Synthesis header. Read
${CLAUDE_PLUGIN_ROOT}/prompts/synthesis.md and write your synthesis UNDER
that header following its structure (Consensus / Divergence / Recommendation)
and calibration rules.
Step 4: Notify User of Saved Output
After displaying the synthesis, tell the user:
(use this emoji 💾) Full output saved to .claude/council-cache/council-TIMESTAMP.md (use the actual filename)
This lets them review the complete responses later.
Output Format: Provider Names
The output file uses emoji prefixes to visually distinguish providers.
Preserve this format when displaying results:
| Provider | Prefix |
|---|
| Gemini | 🟦 Gemini |
| OpenAI | 🔳 OpenAI |
| Grok | 🟥 Grok |
| Perplexity | 🟩 Perplexity |
