slop-grader
CLI and agent skill that checks every line of a document against every rule with Jev and outputs flags for an AI agent to fix.
Overview
slop-grader is a rule-based grader for text files powered by Jev. It runs every rule against every line independently and in parallel, so nothing is skimmed, and then you paste its output to an AI agent, which separates real violations from false positives and proposes concrete replacements. Built-in rulesets cover AI slop (for example banned words, binary contrast, faux insight and colon reveals), grammar and technical documentation quality, and rulesets are plain Markdown that can be validated offline with --check.
Line rules are evaluated with ten lines of context either side, document-scope rules rate the whole text on qualitative rubrics, and dynamic batching groups lines up to token limits. Incremental line caching re-evaluates only edited lines, so unchanged text costs no API calls. Because Jev returns typed probabilities rather than generated text, the README says thousands of checks cost cents. It works with TypeSafe or OpenRouter keys and is published on npm.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Does this line violate rule X (e.g. binary_contrast, banned_word)? | — |
| Score | How well does the whole document meet this qualitative rubric? | — |
How it works
- Run slop-grader with a ruleset on a document
- Every rule is asked about every line with surrounding context
- Output line-level flags and document scores
- Paste into an AI agent to plan and apply fixes
Strengths
- Exhaustive per-line checking
- Cache makes re-runs nearly free
- Rules written in Markdown
Caveats
- Flags need agent or human review for false positives
- Text is sent to the provider
Getting started
export TYPESAFE_API_KEY=..., then npx @lukstei/slop-grader@latest -r no-ai-slop examples/slop.md.At a glance
- Type
- Agent skill
- Status
- beta
- Language
- TypeScript
- Runtime
- Node CLI + agent skill
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Content quality: CLI tool that grades text files against custom rulesets for AI slop, grammar, and technical doc quality using Jev scores and line-level flags, then guides an AI agent to auto-fix violations.
yibie/awesome-jevRule-based CLI and agent skill that grades text against custom rulesets for AI slop, grammar, and technical documentation quality, and guides an AI agent to auto-fix violations
AnotiaWang/awesome-decision-modelsRule-based CLI and agent skill that evaluates text and markdown files against custom rulesets for AI slop, grammar, and technical doc quality using Jev scores and line-by-line violation flags, then guides an AI agent to auto-fix violations.
logicrw/awesome-jev-projectsRule-based text grader that uses Jev scores and line-by-line flags to audit documents against custom rulesets and guide an AI agent to auto-fix violations.
cobanov/awesome-jevRule-based slop grader for text files powered by Jev: evaluates markdown against customizable rulesets, outputting violation flags and scores for an AI agent to auto-fix.
hellogumbo/awesome-jevGrades prose against twenty-one named writing tics, asking every rule about every line, and writes its findings as a brief for a coding agent to act on.
MrJev/awesome-jev
Frequently asked questions
What does Jev decide?
Whether each line breaks each rule, plus document-level rubric scores.
How is it different from asking an LLM to check text?
It checks every rule on every line separately with typed probabilities instead of a model skimming.
Are re-runs expensive?
No; only edited lines are re-evaluated.
Related projects
Node tool that crosschecks a folder of Markdown notes for contradictions and repeats, and ranks notes by any question, using Jev.
ProjectbetaJavaScriptClaude Code plugin that logs your prompts locally and later has Jev score how well you prompt the coding agent over time.
PluginbetaTypeScriptChrome extension that grades readable page sections on ten rubric scores with Jev via Vercel AI Gateway and gives the page an A-E grade.
ProjectbetaTypeScriptTournament engine that ranks texts with Swiss pairing, Elo and TrueSkill, using Jev or local models to judge each head-to-head duel.
ProjectprototypePythonCLI that transcribes a video, has Jev score every sentence against rubric flags, and renders a 16:9 edit with a live meter.
AppbetaPythonFree web tool that checks any URL for 35 tells of AI-generated design and returns a slop score in about 250 ms.
Demoproduction