yzhao062/agent-style

21 writing rules for AI coding and writing agents. Drop-in for Claude Code, Codex, Copilot, Cursor, and Aider, so their output reads like a tech pro.

What it solves

It addresses the tendency of AI agents to produce prose with "AI tells"—such as overusing transition words, excessive bullet points, and passive voice—which often lacks the precision and clarity of professional technical writing. It provides a structured way to make AI agents write like technical professionals by applying established writing standards.

How it works

The project provides a curated set of 21 English writing rules (12 canonical rules from authorities like Strunk & White and Orwell, and 9 field-observed rules targeting common LLM failure modes). These rules are integrated into AI agents in two ways:

  1. Soft Enforcement: The rules are loaded into the agent's system prompt or project configuration at generation time, nudging the model to follow them during the first draft.
  2. Skill-based Review: An optional second-pass audit (style-review) that analyzes a draft against the rules and generates a polished version.

Who it’s for

  • AI coding and writing agents (e.g., Claude Code, Cursor, GitHub Copilot, Aider, Codex) and their users.
  • Maintainers of agent configurations who want generated technical prose to follow literature-backed conventions rather than generic LLM patterns.

Highlights

  • Dual Rule Set: Combines classic writing guides with rules derived from observing LLM outputs from 2022-2026.
  • Broad Integration: Supports a wide array of adapters for popular AI tools, including Claude Code, Cursor, and Copilot.
  • Mechanical Validation: Includes a sanity benchmark showing significant reductions in "AI-tell" violations across multiple flagship models.
  • Technical Scope: Specifically optimized for API docs, design docs, research papers, and READMEs, while explicitly excluding fiction and marketing copy.

Related

  • Dispatch
    SimpleEnglish Agent Skill Enables LLMs to Write in ASD‑STE100 Simplified Technical English

    The SimpleEnglish agent skill forces LLMs to produce documentation that complies with the ASD‑STE100 standard, cutting STE violations by 72.9% and shortening output across multiple Claude models.

  • Project
    addyosmani/agent-skills

    A library of production-grade engineering workflows and quality gates for AI coding agents to ensure they follow senior-level software development best practices.

  • Project
    msitarzewski/agency-agents

    The repository provides a library of markdown‑based AI‑agent prompts (e.g., Front‑end Developer, AI Engineer, Prompt Engineer) and scripts to install them into various LLM‑powered coding assistants such as Claude Code, Cursor, Gemini CLI, etc. A companion desktop app automates installation. It’s a prompt‑template collection, not a model‑training framework.

  • Dispatch
    Improving LLM-Assisted Code Quality with agent.md

    A developer shares a system for using an agent.md file to enforce strict coding standards and commit message rules for AI agents, reducing the need for repetitive manual reviews.

  • Project
    iamfakeguru/agent-md

    agent‑md is a lightweight framework that installs rule files, git‑hooks, and helper scripts to turn AI coding assistants (Claude Code, Codex, Cursor, Windsurf) into contract‑driven agents. It provides a single source‑of‑truth markdown file (AGENT.md) plus enforceable checks (type‑check, lint, tests, visual screenshots, state updates) that block the model until the required evidence is present.