JasonColapietro/suede-creator-skills / skills/suede-code
suede-code
VerifiedUse when explicitly asked "review this PR" or to grade a diff, branch, or commit range. Suede returns evidence-backed P0-P3 findings, fix paths and an A-F ship grade; supports findings-only, grade-only and fix modes, seven review lanes, security checks, deploy safety and commit dirt. NOT FOR: automatic CI enforcement (use suede-ci-gate), agent evaluations (use suede-ai-eval), implementation-plan search (use suede-graph-flo-xr), or parallel fix coordination (use suede-agent-teams).
| File | Kind | Size |
|---|---|---|
| CARD.md | markdown | 4.4 KB |
| SKILL.md | skill md | 18.8 KB |
| agents/openai.yaml | config | 433 B |
| references/grading-contract.md | markdown | 3.1 KB |
| references/maintainability-baseline.md | markdown | 3.4 KB |
| references/security-baselines.md | markdown | 3.3 KB |
| references/threat-verify.md | markdown | 1.4 KB |
| references/traps-swift-ios.md | markdown | 3.7 KB |
| references/traps-web.md | markdown | 10.7 KB |
| references/usage-contract.md | markdown | 842 B |
| references/worked-example-grade-only.md | markdown | 7.9 KB |
| references/worked-example.md | markdown | 11.3 KB |
| scripts/commit-dirt-scan.shexecutable | script | 3.3 KB |
references/maintainability-baseline.md
View on GitHub# Suede Maintainability Baseline
Twelve structural checks the review carries even when a repo documents no
coding standards. Load on `--standard` and `--deep`. Match them against the
diff only: flag what the change introduces or makes worse, never the whole file.
## Rules that travel with the baseline
1. **The repo wins.** A documented project rule outranks this list. If the repo
endorses a pattern a check below would flag, stay silent.
2. **Label it as judgment.** Every finding here reads "possible <check name>",
never as a hard violation. Default severity is P3. Raise to P2 only when the
pattern sits on a changed critical path and you can name how it raises
defect risk there. Confidence is `medium` at most, because these are read,
not executed.
3. **Leave tooling's job to tooling.** If a linter, formatter, or type checker
already catches it, do not report it by hand.
4. **Stay inside the diff.** Pre-existing patterns outside the changed lines
are not findings. At most, add one "residual risk" line when the change
builds on top of one.
## The twelve checks
Each entry names the pattern, then the usual repair.
- **Name hides the job.** A function, variable, or type whose name does not
say what it does or holds. Rename it. If no accurate name comes, report the
muddy design instead.
- **Logic pasted twice.** The same shape of logic appears in more than one hunk
or file in this change. Extract it once and call it from both places.
- **Method lives in the wrong place.** A method that reads another object's
fields more than its own. Move it next to the data it keeps reaching for.
- **Fields that always travel together.** The same three or four values move
through several signatures side by side. Give them one type and pass that.
- **String standing in for a concept.** A raw string or number carries a domain
idea (currency, status, ID kind) that deserves its own small type.
- **Same branch ladder in several places.** An identical `switch` or `if` chain
on the same value recurs across the change. Replace with one lookup map both
sites share, or with polymorphism.
- **One change, many scattered edits.** A single logical change forces small
edits across many files. Gather what changes together into one module.
- **One file, many unrelated reasons.** A module edited for several unrelated
purposes in the same diff. Split it so each module changes for one reason.
- **Built for a future nobody asked for.** Parameters, hooks, or abstraction
layers the spec does not need. Delete them and inline until a real need shows.
- **Long navigation chains.** Callers walk `a.b().c().d()` through internals
they should not depend on. Hide the walk behind one method on the first object.
- **Pass-through layer.** A class or function that mostly forwards calls. Cut
it and call the real target.
- **Inheritance mostly ignored.** A subclass or implementer that overrides or
discards most of what it inherits. Drop the inheritance and compose instead.
## Output shape
Report as a normal P2 or P3 finding, with the check named and the hunk cited:
```
[P3] lib/orders.ts:88, possible Fields that always travel together: (currency,
amount, region) pass through 3 signatures in this diff; bundle as a Money type.
Confidence: medium
```
These findings feed the Technical Debt read inside the grade. They never
outnumber or outrank correctness findings, and they never move the ship gate
unless rule 2's P2 condition holds.
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: 471k | 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: 471k | 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: 471k | 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 |