Review class

Advise only. Reads changes, reports findings, and never writes a file or commits. Shared rules: CONVENTIONS.md.

11 skills

🔍

/claude-design-pull

snapshot→review→verdict

What it does

ALSO use it — before writing any more UI code — whenever the user says the design has NOT landed: \"I still see the old design\", \"I don't see the design changes\", \"none of the…

Agents

design-parity-checker

one `design-parity-checker` per page, in parallel — 3 pages is 3 agent calls · v0.1.0

🔍

/dead-code-finder

snapshot→review→verdict

What it does

Detects the repo's own dead-code tooling at runtime (ts-prune/knip, vulture, deadcode, cargo-udeps) and degrades to a manual export-vs-import cross-reference; reports candidates with a…

Agents

dead-code-finder

1–2 agent calls · v0.2.0

🔍

/deps-upgrade

snapshot→review→verdict

What it does

Surveys the WHOLE current dependency manifest for available upgrades, flags each with a semver risk class and breaking-change signals, and proposes a prioritized upgrade plan — it…

Agents

dependency-reviewer · deps-upgrade

1–2 agent calls · v0.1.0

🔍

/pre-push-review

snapshot→review→verdict

What it does

Runs up to five review dimensions (secrets, correctness, tests/build, dependencies, style) via a Workflow-tool pipeline — the secret scan first as a gate, then the rest in parallel —…

Agents

correctness-reviewer · dependency-reviewer · review-report-writer · secrets-reviewer · style-reviewer · tests-build-runner

up to 5 dimension agents in parallel after the secret-scan gate, plus 1 report-writer after aggregation (Step 5.5) — so up to 6 · v0.5.0

🔍

/review-correctness

snapshot→review→verdict

What it does

Reviews the changed lines and their blast radius for logic errors, regressions, unhandled edge cases, and missing validation.

Agents

correctness-reviewer

1–2 agent calls · v0.3.0

🔍

/review-dependencies

snapshot→review→verdict

What it does

Flags added/removed/upgraded dependencies, license changes, and supply-chain risk signals in the diff.

Agents

dependency-reviewer

1–2 agent calls · v0.3.0

🔍

/review-secrets

snapshot→review→verdict

What it does

REQUIRES a dedicated secret scanner (gitleaks/trufflehog/detect-secrets) on PATH — runs it over the diff first as a HARD GATE (BLOCKs with install instructions if none is installed) —…

Agents

secrets-reviewer

1–2 agent calls · v0.4.0

🔍

/review-style

snapshot→review→verdict

What it does

Checks the diff against surrounding code conventions, commit-message hygiene, and leftover debug/TODO/console output.

Agents

style-reviewer

1–2 agent calls · v0.3.0

🔍

/review-tests-build

snapshot→review→verdict

What it does

Auto-detects the right commands for whatever stack the repo uses (Node, Python, Go, Rust, JVM, Make/just, or commands documented in CLAUDE.md/AGENTS.md) — never hardcodes one project's…

Agents

tests-build-runner

1–2 agent calls · v0.3.0

🔍

/security-deep-review

snapshot→review→verdict

What it does

Runs the same mandatory secret-scanner gate first, then a Workflow-tool pipeline — parallel specialist finders sweep independent lenses (injection/authz…

Agents

security-finder · security-verifier

typically 5 finder calls + up to 15 verifier calls (commonly 15–30 total agent calls for a mid-size diff) · v0.1.0

🔍

/test-gap-finder

snapshot→review→verdict

What it does

Detects the repo's own test/coverage tooling at runtime (jest/vitest --coverage, pytest-cov, go test -cover, cargo tarpaulin) and, when none is configured, degrades to a static…

Agents

test-gap-finder · tests-build-runner

1–2 agent calls · v0.1.0