Skip to content

gabimoncha/oh-my-t3 / skills/omt-code-review

omt-code-review

Verified

Code review with subagents in T3 Code. Use when the user asks to review uncommitted changes, a branch, a commit, or a pull request. Accepts --profile. Runs report-only reviewers and an oracle, merges the findings, and offers fixes.

FileKindSize
SKILL.mdskill md7.3 KB
pr-extras.mdmarkdown1.8 KB

pr-extras.md

View on GitHub
# PR review extras

Run each extra whose switch is not `off`, at the PR step it names, in the PR checkout when it runs code. Save files under `omt paths --evidence`, and put each result in the PR step 5 report.

## CI (`review.ci`)

At PR step 1, read `gh pr checks <n> --json name,state,bucket,link,workflow`, and save the last 400 lines of `gh run view --job <job-id> --log-failed` per failing check as `ci-<slug>-<check>.log`. With `review.ci: fix`, fix each failure after PR step 4, run its local command again, and send the round's CI fixes to one fix review, up to `maxReviewRounds`.

Never wait with `--watch`: when checks are pending at PR step 5, call `watch_pull_request` last, and `unwatch_pull_request` when the review ends.

## Coverage and CRAP (`review.crap: report`)

Before PR step 3, run the coverage command once as `omt lock run coverage -- <command>`, with lcov output to `coverage-<slug>.lcov`. Each reviewer computes `CRAP = complexity² × (1 − coverage)³ + complexity` per added or changed function, with McCabe complexity, and coverage 0 without a coverage file. Report each function above 30 as P2, with its three values.

## Changed-flow QA (`review.verify: changed-flows`)

At PR step 3, when the PR changes a screen, a CLI command, or an API route, dispatch one `qa` child with the changed flows, the build and run commands, and the changed paths. It changes no files and names the test file that covers each flow, or `none`. Each `NOT VERIFIED` flow becomes a P1 finding.

## Test offers (`review.tests: offer`)

At PR step 6, offer one e2e test per verified flow that no test covers, with its file path. After the owner's yes, write the tests yourself in the main thread, and dispatch one `qa` child with `mode: "wait"` to run them. Send the tests through one fix review before the commit.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall