What problem does it solve?
This Skill helps you tune DOS enforcement policy using real outcome evidence so you can reduce false denials without weakening valuable catches.
Core Features & Use Cases
- Outcome-driven policy tuning: Adjust intervention policy knobs based on false-deny and held-catch evidence instead of guesswork.
- Safe iterative evaluation: Compare candidate changes against a baseline using measured net_task_delta and only keep proven improvements.
- Guardrailed enforcement changes: Work with intervention policy settings, intervention ladder ranks, and improvement thresholds while avoiding runtime-logic edits.
- Use Case: A maintainer wants to loosen an overly strict policy that is blocking legitimate work, but only if the new settings preserve catch quality and pass the kernel’s checks.
Quick Start
Ask the assistant to review DOS enforcement outcomes, tune the policy knobs in dos.toml, and keep only changes that measurably improve net_task_delta while preserving suite health and truth cleanliness.