Run adversarial multi-agent code reviews

adversarial-workflowskillsetup L20
dungnthe130669/adversarial-workflow
What it does

Run multi-agent attack-verify workflows

Best for

When you want battle-tested output that survives independent scrutiny rather than single-agent rubber-stamping.

Inputs
  • · User request (bugfix, feature, security, test, etc.)
Outputs
  • · Verified solutions
  • · REVIEW.md
  • · Priority-ranked findings
Requires
  • · Claude Code
  • · Node.js >= 16
Preconditions

Claude Code session active, workflows installed via npx

Failure modes

No consensus among verifiers, timeout on large scope

Trust signals
  • · 11 specialized workflows
  • · Parallel attacker pool
  • · Verifier gate
  • · CVE/debt scanning