
Octen Mcp
Give Claude Code or Cursor a registered MCP tool that turns any public URL into structured, LLM-ready markdown so you can research competitors, docs, and landing pages inside the agent without copy-pa
Overview
io.github.Octen-Team/octen-mcp is a Build-phase-adjacent MCP server for the Idea research shelf that converts URLs into LLM-ready markdown with page category, structure, and query-driven highlights via the Octen API.
What is this MCP server?
- Fetches a URL through Octen and returns LLM-ready markdown instead of raw HTML noise
- Enriches each page with category, page_structure, and query-driven highlights for faster agent reasoning
- Ships as npm package octen-mcp (v0.1.5) with stdio MCP transport for Claude Code–style hosts
- Authenticates with required OCTEN_API_KEY from https://octen.ai (self-serve signup)
- Optional OCTEN_API_URL override when you need a non-default https://api.octen.ai base
- Server version 0.1.5 per MCP server.schema.json
- Single required secret environment variable: OCTEN_API_KEY
- Default API base OCTEN_API_URL=https://api.octen.ai (overridable)
Community signal: 10 GitHub stars.
What problem does it solve?
Agents stall when you need accurate text from a live URL but only have messy HTML, paywalled snippets, or hours lost copying pages by hand.
Who is it for?
Solo builders running Claude Code, Cursor, or Codex who repeatedly feed competitor sites, docs, or articles into planning chats and want consistent markdown plus structure metadata.
Skip if: Teams that need full browser automation, authenticated scraping behind corporate SSO, or offline crawling at scale without an Octen API key and network access.
What do I get? / Deliverables
After you register octen-mcp and set OCTEN_API_KEY, your agent can pull clean markdown and structured page signals from URLs on demand during research and build workflows.
- LLM-ready markdown extracted from a submitted URL
- Per-page category and page_structure metadata for agent navigation
- Query-driven highlights summarizing what matters on the fetched page
Recommended MCP Servers
Journey fit
Solo builders most often wire URL-to-markdown fetchers at the start of a project when they are still gathering signals from the web before committing to a build. Research is the canonical shelf because the server’s value is ingesting live pages—competitors, docs, articles—into agent context with category, structure, and highlights rather than operating production systems.
How it compares
Octen MCP integration for URL-to-markdown ingestion, not a Playwright-style browser skill or a curated skills marketplace.
Common Questions / FAQ
Who is io.github.Octen-Team/octen-mcp for?
It is for indie and solo builders who use MCP-capable coding agents and want URLs turned into LLM-ready markdown with extra page structure and highlights through Octen.
When should I use io.github.Octen-Team/octen-mcp?
Use it during research or anytime you need fresh web page content inside an agent session—for competitor scans, reading product docs, or grounding answers on a specific URL.
How do I add io.github.Octen-Team/octen-mcp to my agent?
Install the npm package octen-mcp (v0.1.5), configure stdio transport in your MCP client, set the required OCTEN_API_KEY from https://octen.ai, then register the server so tools can call the Octen API.