Unslop: UI, Text & Code
Three research-backed skills that spot the tells of AI-generated interfaces, writing, and code, plus scanners that can block a workflow until they are fixed.
Interface build and audit guidance with a design-tell scanner
Prose review guidance backed by a companion language study
Source-code review guidance and machine-checkable artifacts
LicenseMIT for code
The code is MIT-licensed. The included public-discussion datasets and excerpts remain attributable to their original authors; read the repository's data notes before reuse.
What this collection gives you
A practical suite for finding the defaults that make AI-assisted work feel interchangeable. Its UI, text, and code skills are derived from reproducible studies of public discussions and can be used while building or as a final audit.
Inside the repository
A practical map of the parts worth opening first.
Unslop UI
Flags common visual defaults and pushes the interface toward choices that are specific to the product and audience.
Unslop Text
Finds recurring AI-writing mannerisms, leftover assistant language, and formulaic cadence without imposing one house voice.
Unslop Code
Catches visible artifacts such as placeholder comments, swallowed errors, generic naming, and chat residue.
Standalone scanners
Run repository checks outside the chat workflow and use their exit codes as a lightweight quality gate.
Best used for
Where this collection earns its place in a working toolkit.
- Auditing an AI-assisted website before launch
- Cleaning marketing, documentation, or editorial prose
- Reviewing generated code for obvious artifacts and weak defaults
- Adding a repeatable anti-template check to CI
Unslop: UI, Text & Code FAQ
Does Unslop automatically make work feel human?
No. It identifies recurring AI-generated patterns and weak defaults. You still need product, editorial, or engineering judgment to decide which findings matter and what a better project-specific choice looks like.
What is the difference between the skills and the scanners?
The skills guide an agent during creation or review. The scanners run repeatable checks outside the conversation and can provide a baseline or a lightweight CI quality gate.
Do I need to use the UI, text, and code skills together?
No. Use only the skill that matches the artifact you are reviewing. Keeping the audit focused makes the findings easier to evaluate and act on.
What can I reuse from the repository?
The code is MIT-licensed. The included research datasets and public-discussion excerpts retain their original attribution requirements, so review the repository's data notes before republishing them.