slop-grader

About slop-grader

slop-grader is an open-source CLI tool that evaluates text documents against custom rulesets. It produces document scores and line-by-line flags, with output designed to be pasted into an AI agent for auto-fixing. Built on the Jev AI model, it checks for issues like AI filler, buzzwords, grammar, and user-defined criteria such as SEO requirements or tone consistency.

Review

slop-grader functions as a programmable linter for prose - it asks structured questions of each line or the whole document and returns flags where rules trigger. The tool doesn't auto-fix text itself; instead, it generates prompts you can feed to a separate AI agent. It runs locally via Node.js but sends text to an external AI server for evaluation.

Key Features

• Rule-based evaluation using yes/no questions applied line-by-line or across an entire document
• Built-in rulesets for English grammar, German grammar, and AI filler detection
• Custom rules authoring via plain language, with a skill file available on GitHub to assist rule creation
• Line-by-line flag output with instructions formatted for direct pasting into an AI agent
• Document-level scoring based on how many rules trigger against the text

Pricing and Value

The tool itself is free and open-source. Running a check requires an account with TypeSafe or OpenRouter; the maker states that evaluating a document costs less than one cent and completes in seconds, due to the underlying Jev model's efficiency.

Pros

• Custom rules adapt to industry-specific terminology, so words flagged as buzzwords in one context can pass in another
• Output integrates directly into AI agent workflows without manual reformatting
• Fast evaluation makes it practical to run checks frequently during drafting
• Rule creation uses plain-language questions, lowering the barrier for non-developers
• Open-source codebase allows inspection and modification of how checks run

Cons

• Flagged lines do not include an explanation of why a rule fired - the user or downstream AI agent must infer the specific problem
• No strictness setting exists to differentiate between a light review and a deep audit of a document
• Not well suited for teams that need an all-in-one tool with built-in fixing capabilities, since slop-grader stops at flagging and relies on a separate agent for corrections

slop-grader fits writers and developers who already use AI agents in their editing workflow and want a fast, repeatable linting step before handing text off for revision. It's most useful where rule sets can be defined clearly - launch copy review, tone consistency checks, or stripping specific overused phrases. Teams looking for a single tool to both detect and correct issues will need to pair it with another service.