kenzik/council

AI Council - Multi-Agent Consultation

Get perspectives from multiple AI agents on the same question.

Prerequisites

At least one of these CLIs installed:

Claude CLI

npm install -g @anthropic-ai/claude-cli
# Set ANTHROPIC_API_KEY

Codex CLI

npm install -g @openai/codex-cli
# Set OPENAI_API_KEY

Gemini CLI

pip install google-generativeai
# Set GEMINI_API_KEY

Individual Agent Commands

Claude

# Ask Claude
claude -p --model opus "Your question here"

# Print mode (non-interactive)
claude --print "Your question"

Codex

# Ask Codex
codex -q "Your question here"

Gemini

# Ask Gemini
gemini -m pro -o text -e "" "Your question here"

Parallel Consultation Pattern

Basic Multi-Agent Query

# Query all agents in parallel
(
  echo "=== Claude ===" && claude --print "Your question" &
  echo "=== Gemini ===" && gemini -m pro -o text -e "" "Your question" &
  echo "=== Codex ===" && codex -q "Your question" &
  wait
)

Structured Comparison Script

#!/bin/bash
QUESTION="$1"

# Query in parallel, save to temp files
claude --print "$QUESTION" > /tmp/claude-response.txt &
gemini -m pro -o text -e "" "$QUESTION" > /tmp/gemini-response.txt &
wait

echo "=== Claude's Response ==="
cat /tmp/claude-response.txt

echo ""
echo "=== Gemini's Response ==="
cat /tmp/gemini-response.txt

# Cleanup
rm -f /tmp/claude-response.txt /tmp/gemini-response.txt

Use Cases

Code Review Perspectives

# Get different review perspectives
CODE=$(cat mycode.ts)

claude --print "Review this code for best practices: $CODE"
gemini -m pro -o text -e "" "Review this code for security issues: $CODE"

Architecture Decision

QUESTION="Should I use microservices or monolith for a startup MVP with 3 developers?"

# Get multiple perspectives
claude --print "$QUESTION"
gemini -m pro -o text -e "" "$QUESTION"

Debugging Approaches

ERROR="TypeError: Cannot read property 'map' of undefined"
CONTEXT="React component fetching API data"

claude --print "Debug this error: $ERROR. Context: $CONTEXT"
gemini -m pro -o text -e "" "Explain and fix: $ERROR in $CONTEXT"

Finding Consensus

After getting responses from multiple agents, ask one to synthesize:

# Save individual responses first, then:
gemini -m pro -o text -e "" "Here are responses from different AI agents about [topic]:

Response 1: [Claude's response]
Response 2: [Gemini's response]

Synthesize these into a consensus view, noting:
1. Where they agree
2. Where they differ
3. The most actionable recommendation"

Role Presets

When querying, you can assign roles:

Software Engineering Council

claude --print "As a software architect: [question]"
gemini -m pro -o text -e "" "As a security engineer: [question]"

Code Review Council

claude --print "Review for maintainability: [code]"
gemini -m pro -o text -e "" "Review for performance: [code]"

Creative Problem Solving

claude --print "Suggest conventional solutions for: [problem]"
gemini -m pro -o text -e "" "Suggest unconventional approaches for: [problem]"

Debate Pattern

Have agents respond to each other:

# Round 1: Initial positions
R1_CLAUDE=$(claude --print "Argue FOR using TypeScript in a new project")
R1_GEMINI=$(gemini -m pro -o text -e "" "Argue AGAINST using TypeScript in a new project")

# Round 2: Rebuttals
R2_CLAUDE=$(claude --print "Respond to this argument against TypeScript: $R1_GEMINI")
R2_GEMINI=$(gemini -m pro -o text -e "" "Respond to this argument for TypeScript: $R1_CLAUDE")

# Synthesis
gemini -m pro -o text -e "" "Synthesize this debate:

FOR TypeScript: $R1_CLAUDE
Rebuttal: $R2_CLAUDE

AGAINST TypeScript: $R1_GEMINI
Rebuttal: $R2_GEMINI

Provide a balanced recommendation."

Best Practices

1. Use same question - Consistent phrasing for fair comparison
2. Run in parallel - Background jobs with & for speed
3. Capture output - Save to files for comparison
4. Note which agent - Label responses clearly
5. Synthesize at end - Use one agent to summarize
6. Match agent to task - Claude for nuance, Gemini for facts, Codex for code