Show HN: Slopsift – a local, graph-backed linter for AI writing (slopsift.dev)

SlopSift has been launched as a local, graph-backed linter designed specifically for AI-generated writing. This innovative tool utilizes quantized ONNX weights that operate seamlessly in both Node and browser WebAssembly environments, emphasizing a "local-first" approach that prioritizes user privacy. Unlike conventional writing tools that rely solely on word lists and basic grammatical structures, SlopSift employs a custom-trained dependency parser to analyze the grammatical relationships within sentences. This allows it to detect not just vague language and repetitive structures, but also patterns that may indicate the use of borrowed content or overly polished filler, providing users with a deeper insight into their writing. The significance of SlopSift for the AI/ML community lies in its unique approach to natural language processing and its focus on maintaining user control without relying on remote servers. By running the entire model and its deterministic rules on-device, SlopSift protects user data while enhancing writing quality. Its capability to flag suspicious writing styles, such as passive voice or the use of common templates, positions it as a valuable tool in the ongoing conversation about the integrity of AI-generated text. Ultimately, SlopSift emphasizes the need for human oversight in AI-driven writing, fostering a nuanced understanding of how AI can assist while preserving authorship authenticity.