Mintplex Labs · MCP · NinjaTrader 8

Connect AnythingLLM to NinjaTrader 8

RAG + agent tooling. Pair local docs with live MCP state — broker statements meets CrossTrade.

See it in action
anythingllm · crosstrade mcp · sim101

AnythingLLM calling CrossTrade MCP tools against a live NinjaTrader 8 account. Run this yourself in 5 minutes →

Why AnythingLLM

Why pair AnythingLLM with CrossTrade MCP

AnythingLLM combines a document RAG layer with agent tooling. Load broker statements, prop firm rules, your own trading playbook, and have the agent compare against live CrossTrade state. MIT-licensed.

Transport anythingllm_mcp_servers.json with type: "streamable" (not streamable-http)
Highlights

What this combination unlocks

RAG + MCP

Ask questions that span local docs and live trading state.

Docker / Desktop

Self-host or run as a native app.

How it works

The connection chain

AnythingLLM
   │  OAuth-authorized tool call
   ▼
CrossTrade hosted MCP server   (https://app.crosstrade.io/v1/api/mcp)
   │
   ▼
CrossTrade Add-On running inside NT8 on your PC
   │
   ▼
NinjaTrader 8

One URL, one consent screen, one read-only first prompt. The MCP server is hosted by CrossTrade; you don't run anything beyond NT8 + the CrossTrade Add-On.

Ready to set up AnythingLLM?

The full step-by-step setup lives on a dedicated page. Verified against Mintplex Labs's current docs.

Open setup instructions →
Example prompts

Copy-paste prompts that work today

Rule check Compare current state in CrossTrade against my trade rules PDF. Am I in violation?
FAQ

Common questions

Is AnythingLLM a good fit for trading?

Yes if you want a RAG layer combined with CrossTrade tool calls. Load broker statements, prop firm rule PDFs, and your own trading playbook into AnythingLLM workspaces, then ask the agent questions that span local docs and live trading state.

Do I really need a RAG-layered chat?

Useful if you have substantial reference docs (firm rules, trade rules, strategy notes) you want the agent to consult. For pure live-state queries ("what are my positions?"), Claude Desktop or LibreChat is simpler. The RAG layer is the differentiator.

AnythingLLM vs LibreChat / Open WebUI?

AnythingLLM has the strongest RAG-plus-agent story (RAG is its raison d'être). LibreChat has per-user OAuth (best for teams). Open WebUI has the strongest Ollama integration. Pick by the feature you value most.

Cost?

Free, MIT-licensed open source. Desktop install is a one-click affair; Docker for multi-user / server deployment. You pay for the LLM provider key.

Mac, Windows, and Linux?

All three. Desktop installer or Docker. Desktop is single-user; Docker is multi-user with admin/manager/default roles.

Why type "streamable" instead of "streamable-http"?

AnythingLLM uses type: "streamable" (no -http suffix) for Streamable HTTP MCP. Different from every other tool. If you omit type entirely with a url, it defaults to SSE. Don't copy configs from Continue or LibreChat.

Does it support OAuth for remote MCP?

Not natively. Use a bearer token in headers. Get it via another OAuth-capable client (Claude Desktop, Cursor) and paste into the anythingllm_mcp_servers.json file.

MCP config disappears on restart in Docker — why?

STORAGE_LOCATION volume isn't properly mounted. The JSON file is auto-created when you first open Agent Skills but gets regenerated empty if the storage volume isn't persisted. Verify the docker run mount is correct.

Connect AnythingLLM to NinjaTrader 8 in 5 minutes

Free Sim101 walkthrough. Read-only first. Upgrade scope only when a real workflow needs it.