Kartik-Doye/Ravengard-AI-Recruiter / .agents/skills/unslop
unslop
VerifiedRemoves accidental complexity, duplication, dead code, abstraction leakage, naming debt, and generated cruft in small behavior-preserving batches. Use when a working repository is hard to change or contains messy AI-generated code.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 10.3 KB |
| references/cleanup-protocol.md | markdown | 4.0 KB |
| references/research.md | markdown | 766 B |
| references/slop-catalog.md | markdown | 3.7 KB |
references/slop-catalog.md
View on GitHub# Slop catalog: what to hunt, where it hides, why it costs
Ranked by how often it wrecks vibe-coded projects. Evidence rule applies
to every entry: file:line or it isn't a finding.
## Correctness-adjacent (usually Critical)
- **Swallowed errors**: empty catch blocks, `.catch(() => {})`,
`except: pass`, errors logged then ignored. The bug you can't see is
the one that ships.
- **Duplicated schemas/models**: the same entity defined in 3 places
(DB schema, API type, frontend interface) already drifting apart. One
change updates one copy; production disagrees.
- **Dead auth/validation paths**: checks that never run (middleware
mounted on a route that bypasses it, client-side-only validation for
money math).
- **Copy-pasted logic with a "fix" in one copy**: the pagination bug
exists twice; someone fixed one instance.
## Structural (usually High leverage)
- **Mega-components / god-files**: a 2,000-line React component or
1,500-line service file holding five responsibilities. Split by the
seams that already exist, not by a new architecture.
- **Duplicated logic**: the same transformation in several files;
consolidate toward one home.
- **Premature / Zombie abstractions**: base classes with one subclass,
strategy patterns with one strategy, "flexible" config for one use. AI
agents often hallucinate these layers from their 2015-era training bias.
Inline them; add abstraction when the second real case appears.
- **Needless indirection**: wrapper-of-wrapper calls, event buses for
two listeners, "managers" that manage nothing. Flatten.
- **Giant utility files**: `utils.ts` as a landfill; split by domain or
dissolve into actual modules.
## Dead weight (usually Medium)
- **Dead code**: unreachable branches, functions nothing calls
(verified by search, not vibes), feature-flag remnants.
- **Abandoned files**: `new_version_final_v2.ts`, half-migrated
modules, `old/` directories that linger.
- **Unused dependencies**: imports gone, packages remain (check the
lockfile truth, and dev-vs-runtime distinction).
- **Stale comments**: describing code that no longer exists; comment
blocks from a previous architecture.
- **Debugging leftovers**: `console.log("HERE")`, commented-out
experiments, hardcoded test tokens (also a `harden` finding).
- **Scattered config**: the same constant in five files; env vars read
in random modules.
## Generated-slop signatures (Medium unless structural)
- **Type escape hatches**: `any`-cascades, `as unknown as X`, `@ts-ignore`
clusters where the types were hard instead of wrong.
- **Inconsistent naming for the same concept**: `user`/`account`/
`profile` interchangeably; pick per the domain, align gradually.
- **Inconsistent patterns**: three data-fetching styles, two state
systems, four ways to show a loading spinner. Standardize on one,
migrate in batches.
- **Card-grid UI slop / decorative junk** → `polish`'s territory; note
it, route it.
- **AI Sprawl (Accretion by Avoidance)**: agents creating new files or parallel utility functions because writing net-new code is easier than understanding and refactoring the existing file in their context window. Consolidate these.
- **Hallucination residue**: references to packages that were never
installed, APIs that don't exist (guarded by `try/catch` so "it
works"), dead endpoints still called.
## The "keep it" list (slop that's load-bearing)
- Quirks other code depends on (a "wrong" sort order the UI assumes).
- Vendored/generated code: flag the convention, don't hand-edit
generated files.
- Migration paths mid-flight (old + new coexisting on purpose).
Record these in the report as **kept: load-bearing**, or the next
well-meaning cleanup will break them.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |