//
Narrative comments
comments/narrative
Comments that narrate what the agent did instead of explaining the code that remains.
AI code quality gate
The linter for AI-generated code.
AI agents ship code that compiles and passes tests. Sloplint points teams to scanaislop, which catches the patterns they leave behind: narrative comments, unsafe casts, swallowed exceptions, hardcoded URLs, fake fallbacks, and production TODOs before they become codebase debt.
$
npx aislop scan
50+
detection rules
7
languages
0
LLMs at runtime
aislop scan ./src
~ npx aislop scan ./src
src/api/client.ts
WARNconfig/hardcoded-url line 18
WARNerrors/swallowed line 44
src/components/ScanView.tsx
WARNcomments/narrative line 29
AUTO removing stale imports
GATE failing PR - 5 issues found
47
needs cleanup
Patterns detected
The scanner looks for concrete agent output patterns, not vague style opinions. These are the kinds of changes that make a codebase look finished while hiding real maintenance risk.
any
Unsafe type casts
types/unsafe-any-cast
TypeScript escapes that silence mismatches instead of fixing the shape of the data.
{}
Swallowed exceptions
exceptions/swallowed
Empty catches and success-like returns that keep failures out of logs and dashboards.
TODO
Production TODOs
stubs/prod-todo
Placeholder work that was meant to be temporary but landed in a production path.
URL
Hardcoded URLs
config/hardcoded-url
API endpoints and service URLs that should move into environment-specific config.
1?
Hidden fallbacks
fallbacks/missing-input
Defaults that turn missing counts, impossible states, or failed diagnostics into safe-looking values.
50+
rules across slop, security, types, and dead code
<1s
typical local scan startup
CI
GitHub Actions quality gate
OSS
open-source CLI on GitHub and npm