code-review-policy

What problem does it solve?

This policy reduces inconsistent or non-actionable LLM-driven code review findings by defining severity levels, evidence rules, pass/fail gates, and escalation criteria so reviews are reproducible and decision-ready.

Core Features & Use Cases

• Standardized severity taxonomy (S0–S3) that differentiates blockers, critical, major, and minor issues for RTL and verification artifacts.
• Pass/fail gating rules that require remediation for high-severity items and conditional acceptance for medium-severity findings.
• Evidence requirements that mandate file/line references, failing command or log snippets, expected versus actual behavior, and replay artifacts where available.
• Escalation guidance for immediate stops on data-corrupting issues and routing repeat critical failures to architecture owners.
• Use case: audit an RTL module and produce a findings table with reproducible evidence and required remediation steps for the team and stakeholders.

Quick Start

Apply the policy to classify each review finding with severity, attach reproducible evidence, and generate the Code Review Report with verdict, findings, required actions, and escalations.