All posts
June 10, 2026 2 min readClaude CodeMCPAI coding agentsintegrations

How to Connect LLMtoMD to Claude Code (MCP Setup)

By The LLMtoMD team

Claude Code is a capable autonomous coding agent — but on a long build it loses the thread of your requirements as context compacts. Connecting LLMtoMD gives it a searchable knowledge base of your FRD, API references, and decisions, pulled on demand over MCP. No more re-pasting the spec.

What you need

  • Claude Code installed (terminal or the VS Code extension).
  • A free LLMtoMD account with a document or two uploaded. Connecting is free on every plan.

Add the server

In your project directory, run:

claude mcp add --transport http llmtomd https://mcp.llmtomd.com/mcp

That registers the server for the current project. Then, inside Claude Code, complete the sign-in:

/mcp

Select llmtomd → Authenticate. Your browser opens to the LLMtoMD consent page — sign in and click Allow. Run /mcp again and it shows as connected, with its tools available.

Tip: if it doesn't appear right away, restart Claude Code in that folder — it reads MCP config on startup.

Scope it the way you want

  • This project only (default): claude mcp add --transport http llmtomd …
  • Every project: add --scope user.
  • Share with your team via the repo: add --scope project (writes a committable .mcp.json), or commit one yourself:
{
  "mcpServers": {
    "llmtomd": { "type": "http", "url": "https://mcp.llmtomd.com/mcp" }
  }
}

Anyone who opens the repo in Claude Code auto-detects it (with a one-time approval) — so the whole team's agent shares the same project memory.

Verify it works

Ask Claude Code:

List my LLMtoMD documents.

Then the real use case:

Using my LLMtoMD documents, summarize the auth requirements before you start.

It answers from your actual FRD, cited — not from a guess that drifts three files later.

Why it matters

A committed .mcp.json next to your FRD is the cleanest version of the "memory layer" idea: every agent on the project reads from one source of truth, stays on-spec across long sessions, and stops burning tokens on re-pasted context. More on that — including the token math — in Give Your AI Coding Agent a Memory.


Wire your spec into your agent. Convert your first document free →, or see all the integrations.

Convert anything to AI-ready Markdown

PDFs, Office docs, images, audio, and whole websites — clean Markdown and RAG-ready exports for your LLM, in seconds.