Model Context Protocol

Give your AI agent a quant-research tool — in 30 seconds

Factor Weave speaks MCP. Drop one block into Claude Desktop, Cursor, or any MCP-aware client and your agent can pull factor data, run similarity search, and read the market regime across 12,000 tickers — in plain language, no glue code.

1 · Get a key

Sign up free (250 calls/day, no card), then mint a long-lived key from Profile → API Access. MCP access starts at the HOBBY tier.

2 · Add one block to your MCP config

For Claude Desktop, edit claude_desktop_config.json:

{
  "mcpServers": {
    "factorweave": {
      "url": "https://factorweave.com/api/mcp",
      "transport": "streamable-http",
      "headers": { "Authorization": "Bearer fw_live_…" }
    }
  }
}

Stateless streamable-HTTP, protocol 2025-03-26 — works with any compliant MCP client. Restart the client and the Factor Weave tools appear.

3 · Just ask

Your agent now answers questions like:

"What volatility regime is the market in right now, and how long has it held?"

"Show me the 20 strongest-momentum tickers that aren't already overbought on RSI."

"Find historical setups whose factor profile looked like NVDA's today, and pull their factor rows."

"Give me AAPL's factor state on 2024-09-15 — I want a point-in-time snapshot for a backtest."

12 tools, one endpoint

Your agent gets factor lookups, similarity search, top-N screens, market-context analytics, per-ticker report cards, risk-cluster tags, 32-D embeddings, forward-return labels and dataset metadata — twelve tools over a single stateless endpoint. The full tool catalog with every input parameter is in the MCP documentation.

A straight word on what this is

Factor Weave is a research substrate — clean, point-in-time, leak-free factor data and similarity tooling. It is not a return-prediction service, and the MCP tools won't pretend to be one. Our own leak-free testing shows factor similarity does not forecast forward returns — so use these tools the honest way: to screen, explore, and assemble research data your agent reasons over. The methodology is public in the research note.

Get a free key & wire it up

Full protocol details in the MCP docs.