2md

name	2md
description	Use `2md` by Traylinx — a data-extraction API that converts any URL, file, or website into clean, AI-ready Markdown. Supports instant URL-prepend, batch, crawl, sitemap, file conversion, and agent skill generation.

2md — Agent Instructions

You have access to 2md by Traylinx, a data-extraction API that converts any URL, file, or website into clean Markdown optimized for AI context windows.

HOW TO USE: Make HTTP requests using your own built-in tools (fetch, read_url_content, requests, run_command with curl, or any HTTP capability you have).

---

Recommended Workflow

When a user asks you to read, extract, or analyze a webpage:

1. Single page? → Use the URL-Prepend GET shortcut (fastest, zero config)
2. JavaScript-heavy SPA? → Add ?method=browser to the URL-Prepend
3. Multiple pages? → Use POST /api/batch with async mode
4. Entire documentation site? → Use the Crawl Prepend or POST /api/crawl
5. Need a sitemap first? → Use POST /api/map to discover pages, then batch convert

---

1. Instant Convert (GET — URL-Prepend)

The fastest method. Prepend the 2md URL to any target:

GET https://2md.traylinx.com/https://example.com/any-page

Query Parameters

Param	Default	Options	Purpose
format	markdown	markdown, json	json returns structured data with metadata
method	auto	auto, native, static, browser	Use browser for JS-heavy SPAs
force	false	true, false	Bypass cache entirely
maxAge	—	seconds (e.g. 3600)	Only serve cached results fresher than N seconds

Examples

https://2md.traylinx.com/https://example.com/docs?format=json
https://2md.traylinx.com/https://react-app.dev?method=browser&format=json
https://2md.traylinx.com/https://old-page.com?force=true

No API key needed for public web pages.

JSON Response Schema

When using ?format=json, you receive:

{
  "success": true,
  "url": "https://example.com/docs",
  "markdown": "# Page Title\n\nExtracted content...",
  "cache": "miss",
  "method": "native",
  "tokens": { "html": 45000, "md": 3200 },
  "metadata": { "title": "Page Title", "description": "..." },
  "quality": { "wordCount": 1250 }
}

Error Response

{
  "success": false,
  "error": "Description of what went wrong"
}

---

2. Instant Crawl (GET — URL-Prepend)

Crawl an entire site via GET:

GET https://2md.traylinx.com/crawl/https://docs.example.com?depth=3&maxPages=50

Param	Default	Purpose
depth	3	How many link levels deep to crawl
maxPages	50	Maximum number of pages to convert
email	—	Email address for completion notification
webhook_url	—	URL to POST results to when done

Returns a job_id for async polling (see Async Jobs below).

---

3. POST API Endpoints

Base URL: https://2md.traylinx.com

All POST endpoints accept Content-Type: application/json.

POST /api/convert — Convert a single URL

{
  "url": "https://example.com/docs",
  "format": "json",
  "method": "auto"
}

Optional fields:

Field	Type	Default	Purpose
downloadImages	bool	true	Download and embed images
frontMatter	bool	true	Include YAML front matter
screenshot	bool	false	Capture a screenshot of the page
waitMs	number	10000	Wait time for page load (browser method)
maxImageSizeMb	number	10	Max image size to download

POST /api/batch — Batch convert multiple URLs

{
  "urls": ["https://example.com/1", "https://example.com/2"],
  "format": "json",
  "async": true
}

POST /api/crawl — Crawl an entire website

{
  "url": "https://example.com",
  "format": "json",
  "async": true,
  "depth": 3,
  "maxPages": 50
}

Optional: email (string), webhook_url (string) for completion notifications.

POST /api/map — Map a website (discover pages without converting)

{
  "url": "https://example.com",
  "format": "json"
}

Returns a list of discovered URLs. Useful for planning which pages to convert.

POST /api/file2md — Convert uploaded files

Multipart form upload. Supports: PDF, DOCX, images, audio, video.

POST /api/file2md
Content-Type: multipart/form-data

file=@document.pdf
apiKey=sk-...
format=json

Requires apiKey for media/vision extraction (images, audio, video) — as a body field, never a query parameter. With format=json the response body is prefixed by ~1 KB of whitespace padding (proxy keep-alive): trim before parsing. The X-Job-Id response header lets you recover a slow result from GET /api/jobs/{id}/result (send the same x-client-id you posted with).

POST /api/agentify — Generate AI Skill Bundles

{
  "url": "https://docs.example.com",
  "format": "json",
  "async": true
}

Generates SKILL.md + llms.txt + reference files for injecting documentation knowledge into autonomous agents.

---

Format Options (all endpoints)

Format	Content-Type	Description
json	application/json	Structured JSON with markdown, metadata, token counts (recommended)
markdown	text/plain	Raw Markdown text only
stream	text/plain (chunked)	NDJSON streaming logs with final __JSON__ payload

---

Async Jobs

Long-running requests (batch, crawl, agentify) return a job_id.

Poll status:  GET https://2md.traylinx.com/api/jobs/{job_id}
Get result:   GET https://2md.traylinx.com/api/jobs/{job_id}/result

Poll every 5-10 seconds. The status field will be running, done, or failed.

---

Important

• Free tier: Unlimited public page conversions, no API key required.
• Media/vision extraction: Requires an API key. Include apiKey as a field in the POST body (multipart or JSON) — a ?apiKey=... query parameter is ignored and the request will 401.
• Rate limits: The API uses tiered rate limiting. If you receive a 429 response, wait and retry.
• Full documentation: docs.traylinx.com
• Agent discovery: 2md.traylinx.com/llms.txt and 2md.traylinx.com/agents.json