What problem does it solve? Running AI code review normally requires configuring an external LLM endpoint for the review tool. This Skill removes that requirement by letting the host agent perform the review itself, while the open-code-review (OCR) CLI handles only deterministic work: selecting reviewable files and resolving review rules. ## Core Features & Use Cases - LLM-free delegation mode: OCR never calls an LLM; it outputs reviewable file lists, git ref metadata, and rule groups as JSON for the host agent to consume. - Flexible review scopes: Review workspace changes, branch ranges (--from/--to), or single commits, with exclusion patterns and business context via --background. - Structured review workflow: Enforces full file coverage with reviewed/skipped accounting, severity-grouped findings, and optional fix application. - Use Case: On a feature branch, run ocr delegate preview --from main --to feature --format json, fetch rules per file, get diffs with git, and produce a severity-classified review report without any external LLM endpoint. ## Quick Start Ask the agent to review the current workspace changes using OCR delegation mode and report findings grouped by severity.
