AgriciDaniel/anti-slop

Detects patterns of AI-generated slop in text using 33 heuristic patterns. Helps writers and developers maintain authentic, high-quality content by flagging common AI writing artifacts.

A toolkit for reviewing AI-generated text, code, and citations. Detects padding, vendor residue, and vague language, and verifies references with deterministic scanners and structural tests.

Scaffolds and operates an Obsidian vault-based brain that detects and repairs AI slop in prose, code, documentation, and agent output, relying on corpus evidence instead of authorship detection. Use it for quality assurance in AI-generated content.

Reviews AI-generated code for quality defects like hallucinated imports, ceremonial comments, and non-functional tests. Runs deterministic scanners and load-bearing tests to report impact-based severity without authorship claims.

Diagnoses prose quality issues like fabricated facts, padding, and stylistic tells, producing a structured findings report with severity and confidence ratings. Used for reviewing drafts, documents, agent output, commit messages, and more.

Repairs sloppy text by applying fixes from a slop-review report, removing padding, vacuous claims, vague attributions, and vendor residue without fabricating new content. Use after a review identifies issues in a draft, documentation, or agent output.

Runs existence and support checks on citations, DOIs, ISBNs, arXiv IDs, URLs, and package imports. Identifies vendor residue and unresolved attributions to ensure document integrity.