Council

• "run a council on this"
• "get multiple perspectives"
• "debate this decision"
• "stress test this plan"
• architecture, product, strategy, debugging, risk, or founder tradeoffs

$ARGUMENTS

First Step

• references/profiles.yaml

  for panel selection

• references/protocol.md

  for orchestration

• references/verdict-template.md

  for final output shape

• references/personas/<member>.md

  only for the members you actually select

Defaults

• Prefer 3 members unless the user asks for a full panel or the problem is unusually ambiguous.
• Default to

  classic

  +

  architecture

  if nothing else is specified.
• Keep the final verdict compact unless the user asks to see the rounds.

Workflow

1. Resolve The Panel

1. explicit

   --members

2. explicit

   --triad

3. explicit

   --profile

4. keyword triad match
5. fallback default

2. Round 1: Independent Analysis

• Run each selected member independently.
• Keep round 1 blind-first: each member sees only the problem statement and their own persona text.
• Ask for a compact standalone analysis that ends with a clear verdict, confidence, and where the member may be wrong.

• If the host supports explicit subagents or forked contexts, use one independent delegate per selected member.
• In Codex, prefer

  spawn_agent

  , then

  send_input

  and

  wait_agent

  .
• In Amp, prefer one

  oracle

  call per selected member.
• In Claude Code, use parallel or forked agent contexts when they are available. If they are not easy to access, keep the protocol in the main session and separate the member outputs clearly.

You are operating as one member of a structured council.

Persona:
{persona}

Problem:
{problem}

Produce a compact standalone analysis.
End with a clear verdict, confidence, and where you may be wrong.
Do not anticipate the other members.

3. Round 2: Cross-Examination

• Share the round 1 outputs with each member.
• Ask each member to:
  • name the position they most disagree with and why
  • name one insight that strengthened their thinking
  • say whether anything changed
  • restate their position after the exchange
• Prefer sequential execution so later responses can react to earlier disagreements.

Here are the other council members' round 1 analyses:

{peer_outputs}

Respond to all of the following:
1. Which member do you most disagree with, and why?
2. Which member strengthened your thinking, and how?
3. What changed, if anything?
4. Restate your position after the exchange.

Keep it compact and engage at least two members by name.

4. Round 3: Final Position

• Ask for a short final stance only.
• No new arguments unless a host limitation forces a condensed fallback.
• Socrates may ask one final question before stating a position.

5. Synthesis

• Use

  references/verdict-template.md

  .
• Default to the final verdict only.
• If the user asks to show rounds, include concise round summaries after the verdict.

Fallback Mode

1. simulate round 1 as clearly separated persona sections
2. simulate round 2 as explicit cross-exam sections
3. simulate round 3 as final positions
4. disclose that you used the single-agent fallback

Guardrails

• Do not force consensus.
• If the panel converges too quickly, run one counterfactual pass.
• Prefer substance over theater: the council should improve the answer, not just decorate it.