Trove Documentation

Trove gives Claude the tools it doesn’t have on its own. Enable hundreds of ready-made tools from the catalog with one click — pull live data Claude can’t otherwise reach (SEC filings, arXiv, FRED, live weather) or take actions like emailing yourself a summary — all through a single connection to Claude on desktop, web, and mobile. Developers get a CLI, an SDK, and a GraphQL API to build and deploy their own tools. (Under the hood each tool runs on Trove’s cloud as an MCP server; the connection speaks MCP, so any MCP client works too.)

What’s available today

Ready-made tools. Enable a toolkit from the catalog and its tools show up in Claude — arXiv, Wikipedia, SEC EDGAR, FRED, live weather, email-yourself-a-summary, and more, each one click. Browse the catalog.
Build and deploy your own. Author a toolkit (manifest.json + server.ts) against @ontrove/mcp and ship it with one command. Toolkit quickstart.
The Trove connection. Add Trove to Claude (or any MCP client) once; you also get the Core tools (trove_*) — search, browse, discover, and save. Connect an AI assistant.
The Trove CLI. trove manages your tools (trove mcp deploy), captures documents, and queries everything over the one GraphQL API. CLI docs.
The GraphQL API. The single data-access layer every client is built on. Build an app.

Get started

Connect an AI assistant Add Trove to Claude Code, Claude Desktop, or Cursor in one step — then everything you enable shows up there.

Enable a tool Turn on a ready-made toolkit from the catalog — arXiv, SEC EDGAR, FRED, email-yourself-a-summary, and more — and its tools appear in the same connection, ready for Claude to use.

Build your own toolkit Expose your own live tools to Claude — build and deploy in about five minutes.

Build an app Query Trove’s GraphQL API directly. Search documents, save content, and manage toolkits programmatically.

Features

Toolkits, no infra to run. Your tools run on Trove’s Cloudflare Workers, authenticated and always-on, reachable from Claude on every device. Browse the open-source catalog in the toolkit repo.
Semantic search across your knowledge base. Find content by meaning, not keywords, with vector embeddings.
Core tools for Claude and Cursor — search, browse, discover, and save without leaving your AI assistant.
GraphQL API. Build apps, dashboards, or workflows on top of your knowledge base.
Per-user data isolation. Every user gets their own database. No shared tables, no user_id columns.

Learn more

Read about how Trove works to understand how documents, sources, and search fit together.