Description

n8n-nodes-tokenmarkdown


An official n8n community node for TokenMarkdown. Strip 90% to 95% of HTML bloat, cookie banners, tracking iframes, and scripts from any webpage or PDF, converting raw web content into pure, LLM-ready Markdown before sending it into your AI workflows.

—

⚡ Why Use TokenMarkdown in n8n?

Sending raw HTML or generic scraping output directly into OpenAI, Anthropic, or Gemini burns massive token budgets and pollutes LLM context windows with junk:

• Slash Token Bills by 85–95%: A standard documentation or blog page is often 200KB–800KB of raw HTML (~50k–200k tokens). TokenMarkdown extracts the semantic core, delivering 2k–5k tokens.
• Eliminate Context Pollution: Prevents hallucinations caused by cookie consent banners, header navigation links, ads, footer disclaimers, and embedded scripts.
• Deterministic AST Extraction: Powered by an Abstract Syntax Tree (AST) parsing engine designed specifically for RAG embeddings and LLM ingestion.
• Zero Runtime Dependencies: The node has zero external runtime dependencies (dependencies: {}), ensuring rapid execution and rock-solid stability in both cloud and self-hosted n8n environments.

—

📦 Installation

Via n8n Community Nodes (Recommended)

1. In your n8n workspace, navigate to Settings > Community Nodes.
2. Click Install.
3. Enter n8n-nodes-tokenmarkdown in the npm package name field.
4. Agree to the risks of installing community packages and click Install.

Via Docker / Self-Hosted CLI

In your n8n root directory (or container custom directory):

npm install n8n-nodes-tokenmarkdown

—

🔑 Authentication & Pricing

TokenMarkdown works immediately with zero configuration:

• Anonymous Free Tier: 50 free extractions out of the box (leave credentials empty).
• Developer Starter Tier: Enter your developer work email in credentials to unlock 100 free requests (no credit card required).
• Production API Key: For production pipelines and high-volume workloads:

* Builder ($29/mo): 2,500 extractions / month. Subscribe here
* Pro ($79/mo): 10,000 extractions / month. Subscribe here
* Agency ($149/mo): 40,000 extractions / month.

To configure credentials in n8n:
1. Go to Credentials > New Credential.
2. Search for TokenMarkdown API.
3. Enter your API Key (or enter your Developer Email for free tier scaling).

—

🚀 Node Operations & Parameters

Operation: Extract Clean Markdown

| Parameter | Type | Required | Description |
| :— | :— | :— | :— |
| Target URL | String | Yes | The URL of the webpage or PDF to extract and convert. |
| Extraction Mode | Options | No | Compression strategy:
• Article Only (Default): Aggressively strips nav/footer chrome for maximum token savings (~90-95%).
• Full Page: Preserves page layout hierarchy while stripping ads/scripts.
• Raw Semantic HTML: Sanitized semantic HTML tags without converting to Markdown. |
| Include Images | Boolean | No | Retain markdown image embeds (!alt). Default: false. |
| Include Links | Boolean | No | Retain markdown hyperlinks (text). Default: true. |
| Include Table of Contents | Boolean | No | Auto-generate a markdown TOC at the top. Default: false. |
| CSS Selector Filter | String | No | Target a specific container (e.g. article.docs-body, #main-content). |

—

📊 Output Data Structure

The node outputs a structured JSON object ready for immediate chaining with AI Agent, Vector Store, or HTTP nodes:

{
  "markdown": "# Stripe API ReferencennThe Stripe API is organized around REST...",
  "title": "Stripe API Reference",
  "excerpt": "Complete reference documentation for the Stripe REST API.",
  "byline": "Stripe Engineering",
  "url": "https://docs.stripe.com/api",
  "tokens": 3420,
  "rawTokens": 78900,
  "reductionPercentage": 95.66,
  "latencyMs": 342
}

—

🛠️ Example Workflows

1. RAG Pipeline: Webhook ➔ TokenMarkdown ➔ Text Splitter ➔ Qdrant / Pinecone Vector Store
2. Autonomous Research Agent: Google Search ➔ TokenMarkdown ➔ AI Agent (GPT-4o / Claude 3.5 Sonnet)
3. Competitor Changelog Tracker: Schedule Trigger ➔ TokenMarkdown ➔ Slack / Discord Notification

—

📄 License & Attribution

• License: MIT
• Author: The Quite Good Project
• Support & Security: [email protected]