council

name	council
disable-model-invocation	true
description	USER-INVOKED ONLY. Do NOT auto-trigger. Convene 3 fresh Codex subagents to give independent opinions on a question, idea, hypothesis, plan, or ambiguity. Use only when the user explicitly invokes this skill ("$council", "/council") or explicitly names it ("run council", "use the council skill", "convoque le council"). Never trigger on musings, opinions, or phrases like "je me demande si" — thinking out loud is not an invocation. If invoked with no extra prompt, infer the current subject from the active conversation and ask clean-context subagents to review it.

Council

Get 3 independent outside views on the current question without dragging subagents through the whole conversation.

Contract

When the user invokes $council with no extra text:

• Treat it as explicit permission to spawn subagents.
• Infer the current problem, decision, plan, code change, or question.
• Do not ask for clarification unless the current subject cannot be identified.
• Spawn clean-context subagents with fork_context: false.
• Brief each subagent with only the neutral problem statement and needed files, commands, URLs, or facts.
• If the user named a model or reasoning level, carry it through. Otherwise use the host defaults.

When the user adds a topic after $council, or asks for 3 subagent opinions in prose, use that as the brief.

Roles

Default to exactly 3 subagents:

• Skeptic: find weak assumptions, risks, missing constraints, and likely failure modes.
• Researcher: verify current or external facts, prioritizing official docs/sources.
• Practitioner: assess practical execution, repo fit, implementation path, testing, and operational tradeoffs.

For codebase, product, strategy, or writing questions, replace these with sharper lenses when that gives better coverage. Keep one lens adversarial, one evidence-focused, and one practical.

Brief

Each subagent prompt must be short and self-contained:

You are one member of a council. Clean context: assume no prior conversation.

Problem:
[neutral description]

Lens:
[skeptic/researcher/practitioner/etc.]

Evidence to inspect:
[paths, commands, URLs, or "none provided"]

Return:
- strongest concerns or recommendations
- supporting evidence
- what would change your mind

Do not leak private or irrelevant conversation history. For repo behavior, inspect files locally first and pass exact paths or commands.

Synthesis

After subagents respond:

• Present a synthesis, not a transcript.
• Separate consensus from disagreement.
• Highlight the strongest challenge, recommended next move, and evidence gaps.
• Do not force agreement when disagreement is useful.