STARTER_CHARACTER = 🕵️‍♂️

Code Review Command

Review a codebase from 9 perspectives against 3 models. The repo is cut into components of at most 25 files, and each of the 27 perspective and model pairs reviews every component, so every file gets all 27 reviews. Each arm is its own opencode run process that works in its own copy of the reviewed commit, navigates it with Serena, and writes its own findings. A sweep per pair then looks for defects the component boundaries hid.

Nothing is packed. There is no orchestrator process and no task fan-out. A reviewer reads the files of the reviewed commit, so a citation points at a real line in a real file. No reviewer reads the live repository.

One ocr scan runs alongside them. It reviews every reviewable file in its own conversation with a generic checklist, which makes it an independent fourth source.

Arguments

If the user provided a path with the invocation, treat it as the target directory relative to the repo root. Otherwise review the whole repo. That is the deliberate default, already accepted by whoever invoked this skill, not a decision to re-confirm.

This skill runs context: fork, dispatched as a subagent with no user present to answer a prompt. Never pause before Step 2 to confirm scope, cost, or model spend. Proceed straight through Step 1 into dispatch. If cost matters, that call was made before invocation, by passing a narrower path.

Models