rlaope/oh-my-hermes / skills/omh-code-review
omh-code-review
Verified[omh] Pull request or changes to vet: bug-first review with evidence. Use when the user says: code-review, review, audit, find bugs, claim audit, evidence audit, README claim, what actually happened.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 13.9 KB |
| references/architecture-delta.md | markdown | 4.9 KB |
| references/finding-validation.md | markdown | 4.9 KB |
| references/native-review-contract.md | markdown | 7.2 KB |
| references/review-dispatch.md | markdown | 11.6 KB |
| references/review-lenses.md | markdown | 5.2 KB |
| references/review-response.md | markdown | 2.4 KB |
| references/risk-selection.md | markdown | 4.3 KB |
| references/smell-baseline.md | markdown | 4.4 KB |
references/smell-baseline.md
View on GitHub# Smell Baseline
The named baseline for maintainability findings. A smell here is a judgement
call to argue from evidence in the diff, never an automatic finding - and the
reviewed repository's own documented standards override this baseline wherever
they conflict. Cite the smell name in the finding so the author can look up the
same definition. Adapted from the classic Fowler/Beck catalog.
## Find the repository's standards first
The override can only fire on standards someone read. Before applying the
baseline, collect the reviewed repository's written conventions: `REVIEW.md`,
`CONTRIBUTING.md`, `AGENTS.md` and `CLAUDE.md`, any `*STANDARDS*.md` (such as
`CODING_STANDARDS.md`), and the linter, formatter, and type-checker
configuration. A dispatched review receives this list under Constraints
(`references/review-dispatch.md`); a review run in place builds it itself. Name
each file found by path and each one looked for and not found, so a reader can
tell a standard that was checked from one that was never opened.
Those files do more than silence smells. A diff that breaks a rule written in
one of them is a finding of its own: cite the file and the rule.
## The twelve baseline smells
| Smell | What it is | The usual fix |
| --- | --- | --- |
| Mysterious name | A name that forces the reader to open the body to learn what it does. | Rename to what it does or returns; a long clear name beats a short opaque one. |
| Duplicated code | The same decision encoded in two places, so one edit needs two. | Extract the shared decision to one owner; leave lookalikes that encode different decisions alone. |
| Feature envy | A function that reads or writes another module's data more than its own. | Move the function to the data it envies, or move the data to the function. |
| Data clumps | The same group of values travelling together through signatures. | Introduce the object the clump is trying to be. |
| Primitive obsession | Domain concepts passed as bare strings/ints so nothing checks them. | Wrap the concept in a type that validates at the boundary. |
| Repeated switches | The same type/kind dispatch re-implemented at several sites. | Centralize the dispatch (polymorphism, a table, one router) so a new case is one edit. |
| Shotgun surgery | One conceptual change that requires edits scattered across many files. | Move the pieces of the concept into one place before the next change. |
| Divergent change | One module edited for many unrelated reasons. | Split the module along its change reasons. |
| Speculative generality | Hooks, parameters, or layers serving only an imagined future caller. | Delete until a real second caller exists. |
| Message chains | `a.b().c().d()` walks a structure the caller should not know. | Hand the caller what it actually needs, or hide the walk behind the owner. |
| Middle man | A layer that only forwards to another layer. | Collapse it; talk to the real owner. |
| Refused bequest | A subtype that stubs, ignores, or overrides most of what it inherits. | Replace the inheritance with composition or split the interface. |
## How to report one
Before recommending shared-code extraction, load `references/architecture-delta.md`
and apply its extraction proof rubric: actual authored callers, compatible
contracts, existing reuse, and net maintenance benefit. A smell is not that proof.
- Name the smell, cite the evidence (`path`, `line_range`, and what shows it),
and say what the fix would be - as a finding, usually `P2`/`P3` unless the
smell hides a correctness risk.
- One instance is a question; a pattern is a finding. Prefer the site where the
next change will hurt.
- Do not report a baseline smell the repository's standards explicitly accept;
cite the standard instead and move on.
- Skip a smell the repository's linter, formatter, or type checker already
enforces on this diff: the rule is switched on and runs in CI or a hook. A
suppression the diff adds (`noqa`, `eslint-disable`, `type: ignore`, an
`allow` attribute) is itself reviewable, because it turns that enforcement
off for exactly these lines.
- New code matching surrounding style beats abstract purity: a smell the whole
file already commits to belongs in a follow-up scope question, not a blocking
finding on this diff.
## Boundary
A smell finding is a maintainability judgement over the diff, not execution,
verification, CI, or merge evidence, and never blocks on its own unless the
reviewed repository's standards say it does.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |