Slopless is a command-line tool for catching what its authors call 'prose slop' in English Markdown documents. It is built on top of textlint and ships more than 50 deterministic rules that look for common patterns of bland, vague, or AI-generated writing. The key word is deterministic: it does not call a language model to make its judgments, it just applies fixed rules and reports findings as structured JSON. The intended workflow is to plug it into an AI writing agent. You install Slopless as a dev dependency in your project, then run a command that installs a 'skill' file for either Codex or Claude Code. Once installed, you tell the writing agent to use the Slopless skill and let it loop: write, run Slopless, read the JSON findings, rewrite, rerun, until the output is clean. The README describes this as the main intended usage pattern. You can also run it directly from the command line against a file path, a glob like docs/**/*.md, or piped stdin. Exit code 0 means the file is clean, 1 means there were findings, and 2 means the command itself failed. The output is always JSON, which the README suggests saving under a .slopless/findings directory with a timestamped filename for later review. For cases where a flagged sentence is intentional, the README shows how to wrap it with textlint-disable and textlint-enable HTML comments to skip specific rules in specific places. The README links out to a wiki for deeper material, including a philosophy page, a comparison against tools like proselint, write-good, alex, and vale, a full rule inventory across seven rule families, and a behavior reference for CLI flags, exit codes, and JSON shape. Slopless is developed by seochecks.ai and the stated goal is to keep written content specific, useful, and recognizably human.
Generated 2026-05-21 · Model: sonnet-4-6 · Verify against the repo before relying on details.