cyberneticlibrary

Multi-lens adversarial engine review

ultra-qa-reviewworkflowsetup L31
robotlearning123/open-dynamic-workflow
What it does

Execute 6-lens QA with per-finding skeptic refutation and synthesis

Best for

Pre-release audit on mature code when adversarial multi-agent review justifies token cost.

Inputs
  • · [object Object]
Outputs
  • · [object Object]
Requires
  • · node
  • · npm/npx vitest
  • · Bash
  • · potentially unsafe code execution
Preconditions

Repo has built dist/; tests are runnable; code is readable

Failure modes

Execution of adversarial code is risk; test suite gaps miss edge cases; 6 lenses can be redundant

Trust signals
  • · Every finding executed to confirm
  • · 3 skeptics per defect
  • · Pre-fixed issues documented