Adversarial Review

2 nodes · 1 podclaude-code, codex

reviewminimaladversarial

Two agents from different runtimes review the same code. Claude Code (Opus) and Codex catch different classes of mistakes. Minimal, focused, immediately useful.

rig up rigs/adversarial-review.yaml

RigSpec

version: "0.2"
name: adversarial-review
pods:
  - id: review
    label: Review
    continuity: ha_pair
    members:
      - id: r1
        agent_ref: "local:agents/reviewer-opus"
        profile: default
        runtime: claude-code
        cwd: .
      - id: r2
        agent_ref: "local:agents/reviewer-codex"
        profile: default
        runtime: codex
        cwd: .
    edges:
      - kind: collaborates_with
        from: r1
        to: r2