# Your AI Writes Better Code When It Knows Your Standards

I've been building something I'm really excited about and wanted to share it here: [CodeContext](https://www.codecontext.app).

## The Problem

If you use AI coding assistants — Claude, Cursor, Windsurf, Copilot — you've probably noticed the same thing I did: they write decent code, but they don't know *your* rules. Your naming conventions, your error handling patterns, which libraries you prefer, how your API responses should be structured. You end up correcting the same things over and over.

[CLAUDE.md](http://CLAUDE.md) files and rules files help, but they live in your repo, get stale, and don't scale across projects or teams.

## What CodeContext Does

CodeContext is a knowledge base for your coding standards and API specs that delivers them directly to AI assistants via [MCP](https://modelcontextprotocol.io) (Model Context Protocol).

Here's the flow:

1.  **You define your standards** — coding conventions, architecture rules, patterns, anti-patterns. Tag them by stack, layer, or team.
    
2.  **You import your API specs** — drop in an OpenAPI/Swagger file and every endpoint gets indexed.
    
3.  **Your AI assistant calls CodeContext** — when it's about to write code, it calls `get_code_context` and gets back only the standards relevant to the current task. No prompt bloat, no irrelevant context.
    

The AI doesn't just get a wall of text — it gets focused, task-specific guidance. Working on a React component? It gets your frontend standards. Writing a database migration? It gets your SQL conventions. Integrating with an external API? It gets the exact endpoints, parameters, and response schemas.

## What Makes It Cool

*   **MCP native** — works with Claude (desktop and CLI), Cursor, Windsurf, VS Code, and any MCP-compatible client. Connect once, standards everywhere.
    
*   **Smart context delivery** — standards are filtered by relevance to the task, not dumped in bulk. Your AI gets signal, not noise.
    
*   **Spaces** — organize standards by project, team, or domain. Scope API keys to specific spaces.
    
*   **API spec management** — import OpenAPI specs and your AI can look up any endpoint on the fly.
    
*   **Team collaboration** — share standards across your team so everyone's AI writes code the same way.
    
*   **Slash commands** — `/codecontext:implement-feature`, `/codecontext:code-review`, `/codecontext:extract-standards` and more, available automatically when connected.
    

## Getting Started

There's a free tier — 25 standards, 3 API specs, 500 MCP calls/month. Enough to try it out and see if it fits your workflow.

1.  Sign up at [codecontext.app](http://codecontext.app)
    
2.  Add your first standards (or use `/codecontext:extract-standards` to have your AI analyze your codebase and generate them for you)
    
3.  Connect your AI assistant via MCP
    
4.  Start coding — your AI now knows your rules
    

Check it out and let me know what you think: [codecontext.app](http://codecontext.app)