slop-grader

Product Introduction

1. Definition: slop-grader is a rule-based command-line interface (CLI) tool and text analysis engine designed for automated content quality assurance. It functions as a static analysis linter for prose, evaluating documents against customizable rulesets to detect low-quality writing patterns, often referred to as "AI slop."
2. Core Value Proposition: It exists to automate the detection of poor writing quality, buzzwords, and structural flaws in text, providing objective scores and actionable feedback. Its primary value is enabling a human-in-the-loop workflow where AI agents can use its precise, line-by-line flags to draft targeted improvements, thereby streamlining content revision and editing processes.

Main Features

1. Customizable, Rule-Based Grading Engine: The tool's core is a flexible rules engine that parses user-defined or built-in rulesets written in Markdown or JSON. It evaluates each line of a text document against "Line Rules" (e.g., checking for banned words, empty adverbs) and the entire document against "Document Rules" (e.g., scoring narrative arc, engagement). It uses the Jev language model (via TypeSafe AI or OpenRouter) to interpret these natural language rules and apply boolean or scoring logic.
2. AI-Agent-Optimized Output: slop-grader generates reports designed for consumption by AI agents like Gemini or Claude. The output includes the original text, line numbers, specific rule violations, and a structured prompt that guides the agent to distinguish false positives and generate concrete replacement text. This creates a tight feedback loop for automated content refinement.
3. Extensible Ruleset Library & Syntax: It ships with built-in rulesets for common scenarios (no-ai-slop, article-scores, tech-docs, grammar-english) and supports creating custom rules for any domain. The rule syntax allows for creative, plain-language questions about any structured or unstructured text, enabling use cases far beyond marketing copy, such as auditing CSV files, legal contracts, git commits, support transcripts, and code reviews.

Problems Solved

1. Pain Point: The proliferation of low-quality, generic, and buzzword-filled "AI slop" in business writing, marketing copy, and technical documentation. Manually identifying and fixing these issues is time-consuming and subjective.
2. Target Audience: Content strategists, marketing managers, technical writers, startup founders, developers reviewing documentation, legal and compliance teams auditing contract language, and customer support leads analyzing agent conversations.
3. Use Cases: Catching AI-generated filler in launch announcements; stripping jargon and puffery from landing page copy; scoring the narrative flow and closing strength of launch emails; auditing git commit messages for adequate context; identifying uncapped indemnity clauses in contracts; flagging hardcoded secrets in code; analyzing support transcripts for unauthorized promises.

Unique Advantages

1. Differentiation: Unlike generic grammar checkers (Grammarly) or style guides, slop-grader is programmable and context-aware. It doesn't just check pre-defined grammar rules; it allows users to define domain-specific quality heuristics. Unlike purely AI-based rewriting tools, it provides transparent, rule-based justification for its flags, enabling controlled, iterative improvement rather than a black-box rewrite.
2. Key Innovation: Its "rules-as-plain-text" approach lowers the barrier to creating sophisticated text analyzers. By combining a simple Markdown rule syntax with a powerful LLM (Jev) for interpretation, it allows non-programmers to encode complex editorial, compliance, or stylistic policies into an executable audit tool, bridging the gap between human editorial judgment and automated analysis.

Frequently Asked Questions (FAQ)

1. What is slop-grader and how does it improve content quality? slop-grader is a CLI tool that scores documents against custom rulesets to flag poor writing patterns like buzzwords and weak structure. It improves content quality by providing objective, line-by-line feedback that guides AI agents or human editors to make precise, actionable revisions, ensuring sharper and more authentic copy.
2. Can I use slop-grader for purposes other than checking marketing copy? Yes, slop-grader's rule-based engine is highly versatile. You can write custom rules to audit CSV data for suspicious transactions, evaluate legal contracts for risk clauses, assess git commit messages for intent, check code for unjustified type assertions, or analyze support chat logs, making it a general-purpose text analysis and policy enforcement tool.
3. How does slop-grader integrate with AI agents for auto-fixing content? The tool outputs a structured report that includes the original text, line numbers, and specific rule violations. This report is formatted as a direct prompt for an AI agent, instructing it to review the flags, ignore false positives, and generate concrete replacement text for each violation, creating an automated draft-fix-review workflow.
4. Do I need an API key to use slop-grader, and which AI models does it support? Yes, slop-grader requires an API key for either TypeSafe AI (using the jev provider) or OpenRouter. It uses the Jev language model by default (jev-latest on TypeSafe, ~typesafe/jev-latest on OpenRouter), but the model can be overridden via the --model flag when using the OpenRouter provider.
5. Is slop-grader suitable for non-technical users like editors or marketers? While the primary interface is a command-line tool, non-technical users can benefit by using pre-built rulesets or collaborating with a developer to create custom rules. The accompanying create-slop-grader-rules AI skill also assists in writing and validating rulesets using natural language, making the rule creation process more accessible.