Contrarian Agents Kit

Agents that deliberately challenge assumptions and play devil's advocate.

Configuration

Scenario

Challenge level

Must provide evidence

Contrarian Agents Dashboard

Challenges

Resolved

Pending

Integration Code

import { createContrarianAgent } from 'agent-tools-kit/cognitive'

const contrarian = createContrarianAgent({
  level: 'moderate',
  requireEvidence: true,
  model: 'gpt-4o',
})

// Challenge a decision with structured feedback
const result = await contrarian.review(decision, {
  assumptions: ['Microservices scale better', 'PostgreSQL for all data'],
  onChallenge: (challenge) => {
    console.log('Challenge:', challenge.assumption, '→', challenge.counter)
  },
  onEvidence: (evidence) => {
    console.log('Evidence:', evidence.source, evidence.citation)
  },
})

console.log('Challenges raised:', result.challenges.length)
console.log('Decision improved:', result.revised)