Machine-readable endpoints for AI assistants and LLM integrations

Overview

AI Toolkit provides machine-readable text files following the llms.txt convention. These endpoints help AI assistants understand and work with the toolkit.

Available Endpoints

/llms.txt — Condensed Reference

A concise overview of all modules with key export signatures. Ideal for including in an LLM's context window when the full reference is too large.

https://ai-toolkit-docs.vercel.app/llms.txt

/llms-full.txt — Complete API Reference

Every exported function, type, interface, and class across all 17 modules with full TypeScript signatures and working code examples.

https://ai-toolkit-docs.vercel.app/llms-full.txt

/prompt.txt — System Prompt

A ready-to-use system prompt for AI assistants that help developers use the toolkit. Includes module overview, import patterns, common patterns, and quick start.

https://ai-toolkit-docs.vercel.app/prompt.txt

Usage

With Claude

Paste the content of /llms-full.txt into your conversation, or reference it as context when asking questions about the toolkit.

With Custom Agents

import { createAI } from '@jamaalbuilds/ai-toolkit/ai';

const response = await fetch('https://ai-toolkit-docs.vercel.app/prompt.txt');
const systemPrompt = await response.text();

const ai = createAI();
const result = await ai.generate('How do I set up vector search?', {
  system: systemPrompt,
});

With MCP

Use the toolkit's MCP module to serve these resources as MCP resources:

import { McpServerBuilder } from '@jamaalbuilds/ai-toolkit/mcp';

const server = new McpServerBuilder({ name: 'toolkit-docs', version: '1.0.0' });
server.defineResource({
  uri: 'docs://ai-toolkit/api-reference',
  name: 'AI Toolkit API Reference',
  handler: async () => {
    const res = await fetch('https://ai-toolkit-docs.vercel.app/llms-full.txt');
    const text = await res.text();
    return {
      contents: [{ uri: 'docs://ai-toolkit/api-reference', mimeType: 'text/plain', text }],
    };
  },
});