Back to directory

antislop

A set of mandatory rules for AI coding agents that filters out generic AI-generated interface designs, copy and code, while leaving room for a project's own design direction to supply the finished look.

antislop

antislop is a set of rules for AI coding agents that removes generic 'AI slop' user-interface designs and writing, without imposing a specific style itself. It is described as a filter rather than a style guide: it prescribes no fixed colors, fonts or layouts, and covers writing and code as well as visual interface work, so AI-generated copy stops reading like generic AI output.

It defines 38 mandatory rules in three tiers, from an absolute hard gate to purpose-gated techniques that require a stated reason and consistency-focused quality locks, plus a toolkit with three adjustable dials for energy, rhythm and motion, and a mandatory pass-or-fail delivery report run before anything ships. It is additive: an agent loads only the specific skills a given task needs. The project notes that a project's own DESIGN.md file is expected to supply visual or stylistic direction; antislop's rules only filter out generic output and cannot invent that direction on their own.

It is intended for use with AI coding agents such as Claude Code.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.