Open-source Slop Mop extension flags low-value posts in LinkedIn feeds

Slop Mop, a free and open-source Chrome extension that flags low-value posts in the LinkedIn feed, was released this week by Tom Frazier, an AI enthusiast, author and business consultant. The code is on GitHub, and the developer says it requires no signup.

Nine tells, three questions

The extension scores posts against nine tells that SEO firm Graphite Growth identified in research into patterns that show up disproportionately often in AI-generated writing: a post that opens with one line and a gap that pushes readers to expand it, heavy hype words, promises that carry no catch, flowery language, a suspiciously neat lesson, engagement bait, empty praise, terms defined by contrast, and stiff phrasing.

Three more questions guard against false positives: whether the writing sounds like a particular person, and whether it contains anything useful to readers — both can only lower the slop score. A twelfth asks how likely a model drafted the post, but it too can only reduce the score, and it cannot get a post flagged on its own, Frazier explains on the extension's website.

Human slop counts too

Frazier calls Slop Mop less an AI detector than a filter for internet garbage, whoever wrote it. "Slop Mop flags poor human content equally to poor AI content," he told The Register. "I feel that is the right decision because AI content can also be good and human content can also be bad."

AI-detection outfit Pangram classified 41 percent of the LinkedIn posts longer than 250 words in its sample as fully AI-generated, a figure that excludes posts their authors used AI to polish. LinkedIn recently added a button letting users report a post as "Seems Like AI Slop." Posts that cross Slop Mop's threshold are flagged or hidden locally, not removed from the platform.

A testbed for Jev

Frazier told The Register the main motivation was not annoyance at the feed but curiosity about Jev, the probabilistic decision-making model from TypeSafe released the week before. Jev is not a chatbot: it returns decisions for other software through an API, and developers have found many uses for it since launch.

Because Slop Mop runs on Frazier's own Jev account, he capped usage at 250 analyzed posts per installation per day, well beyond typical LinkedIn use. Early installs cost roughly $0.005 per user per day. "If this project goes viral, I might re-evaluate the free service portion of it but for now I am happy to eat those costs to understand tuning parameters and build the dataset." In his view the project shows off "a new class of AI infrastructure that is here to stay"; context-aware decisions with a confidence score, he predicts, "will do more for enterprise workflows than LLMs have to date."