What It Does
Compares paired agent-harness task trials (same tasks run on two or more harnesses) from redacted, supplied trial records, without treating the supplied success/duration claims as independently verified.
Use Cases
- Teams preparing review-ready evidence before operational handoff.
- Maintainers who need deterministic local checks without contacting live systems.
- Operators documenting risks, gaps, owners, and next actions for release governance.
Requirements
- Python 3 standard library only.
- No API keys required.
- OpenClaw-compatible local skill workspace with redacted evidence notes.
Price
$4.99 one-time purchase.