@easydocs/core

Core engine for EasyDocs — turns your API's real traffic into OpenAPI specs, with storage and the capture queue. Local-first and self-hostable; an AI model does the generation, with an offline mode (Ollama).

You don't need to install this directly. Framework adapters (@easydocs/express, @easydocs/fastify, etc.) depend on it automatically.

What it does

• Captures HTTP traffic from middleware adapters
• Runs a background queue so captures never block requests
• Detects response shape changes and only re-runs AI when needed (sampling)
• Generates OpenAPI 3.0 Operation objects via OpenAI, Anthropic, DeepSeek, or Ollama
• Stores specs in SQLite (default) or Postgres
• Detects auth schemes from request headers and documents them

Direct usage

import { createCapturer, parseConfig, buildCaptureEvent } from '@easydocs/core'

// At setup time — creates a Capturer with its own storage adapter and queue
const capturer = createCapturer(parseConfig({ project: 'my-api', ai: { provider: 'openai' } }))

// Per request — fire and forget, never await
capturer.capture(buildCaptureEvent({
  method: 'GET',
  path: '/users',
  query: { page: '1' },
  params: {},
  requestBody: null,
  responseBody: { data: [], total: 0 },
  status: 200,
  requestHeaders: {},
  responseHeaders: {},
  durationMs: 12,
}))

Configuration

{
  project: 'my-api',           // project slug, default: 'default'
  ai: {
    provider: 'openai',        // 'openai' | 'anthropic' | 'deepseek' | 'ollama'
    model: 'gpt-4o',
    apiKey: '...',             // falls back to OPENAI_API_KEY / ANTHROPIC_API_KEY / DEEPSEEK_API_KEY
    baseUrl: '...',            // for custom OpenAI-compatible endpoints
  },
  storage: {
    type: 'sqlite',            // 'sqlite' | 'postgres'
    url: 'file:./docs.sqlite', // default: ~/.easydocs/db.sqlite
  },
  capture: {
    ignoreRoutes: ['/health'],
    includePaths: ['/api'],
    maxBodySize: 10_000,
  },
  dashboard: {
    autoStart: true,           // spawn dashboard on first capture (dev only)
    port: 4999,
  },
}

AI provider detection

Auto-detection precedence (when no explicit provider is set):

| Environment variable | Provider used | |----------------------|---------------| | ANTHROPIC_API_KEY | Anthropic Claude | | DEEPSEEK_API_KEY | DeepSeek | | OPENAI_API_KEY | OpenAI GPT-4o | | none | Ollama at localhost:11434 (fully offline) |