docs — Agent-First Language Documentation

A modular, standalone repository of programming language and platform documentation, optimized for consumption by AI coding assistants (LLMs) via Neuledge.

Why This Exists

LLMs are only as good as the docs they have access to. Training data is stale, context windows are limited, and hallucinated APIs waste time. This repo provides ground truth — accurate, versioned, concise reference docs that an agent can query in milliseconds.

Structure

languages/          # 28 programming languages (JS, Python, Rust, Go, Java, C#, …)
frameworks/         # 34 web frameworks & libraries (React, Vue, Next.js, Nuxt, Hono, ComfyUI, Cypress, Playwright, Jest, Vitest, Zustand, …)
databases/          # 8 databases & ORMs (PostgreSQL, MongoDB, Redis, Prisma, Vector DBs, …)
devops/             # 8 DevOps tools (Docker, Kubernetes, Terraform, Nginx, Context, Graph, …)
web-platform/       # HTML, CSS, Web APIs
concepts/           # Cross-cutting topics (HTTP, Git, Testing, Databases …)
_template/          # Copy this to add a new domain
scripts/            # Setup and build helpers

Every domain is self-contained — add or remove a directory without affecting anything else.

Quick Start

1. Install Neuledge

npm install -g @neuledge/context

2. Index This Repo

context add .

This parses all Markdown files and builds a portable neuledge.db with full-text search.

3. Start the MCP Server

context mcp

4. Connect Your Agent

Claude Code:

claude mcp add docs -- npx @neuledge/context mcp

Cursor / VS Code — add to settings.json:

{
  "mcpServers": {
    "docs": { "command": "npx", "args": ["@neuledge/context", "mcp"] }
  }
}

Your AI assistant now has instant access to all docs in this repo.

Adding a New Domain

cp -r _template languages/rust    # or any domain name
# Edit the README and add .md files
context add .                      # re-index

Conventions

• One file per topic — keeps context windows small and relevant
• Markdown only — what Neuledge indexes best
• Reference style — concise, factual, no tutorials or marketing fluff
• Code examples on every concept — LLMs learn best from examples

License

MIT