Anyquery
Run SQL over files, databases, and 40+ apps (GitHub, Notion, Chrome, Todoist) — and expose them to LLMs via MCP.
Add to your client
Copy the config for your MCP client and paste it into its config file.
brew install anyqueryPaste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"anyquery": {
"command": "path/to/anyquery/that/you/copied/from/step-3",
"args": [
"mcp",
"--stdio"
]
}
}
}Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf
Before you start
- Install the anyquery binary (Homebrew, APT, YUM/DNF, Scoop, Winget, Chocolatey, AUR, or release binary)
- An MCP-compatible client (e.g. Claude Desktop, Cursor, Windsurf)
- Optionally install plugins from the Anyquery registry to query specific apps/data sources
About Anyquery
Anyquery turns files, databases, and SaaS apps into SQL-queryable tables via SQLite virtual tables and a plugin registry, then exposes them to MCP-compatible LLM clients. After installing the anyquery binary, you point your MCP client at anyquery mcp --stdio; the model can then list tables, read their schemas, and run SQL across everything you've connected. It also offers an HTTP/SSE transport and a MySQL-server mode for traditional BI clients.
Tools & capabilities (3)
listTablesLists all the tables available. When the user requests data or wants an action (insert/update/delete), the model calls this to check if a table corresponds to the request.
describeTableDescribes a table — returns its columns, the supported operations, and SQL examples. Must be called for each table before executeQuery.
executeQueryExecutes a SQL query (SELECT, INSERT, UPDATE, or DELETE) on the database, according to what each table's schema supports.
What this server can do
Anyquery provides tools for these capabilities — tap one to see every MCP server that does the same:
When to use it
- Let Claude or another LLM query your local CSV/JSON/Parquet files with SQL
- Ask an LLM to search and analyze data across connected apps (Notion, GitHub, Chrome history, Todoist, etc.)
- Join data from multiple sources (files + databases + apps) in a single SQL query
- Use Anyquery as a MySQL-compatible backend for BI tools like TablePlus or Metabase
Security notes
Runs locally and queries your connected data sources, files, and app integrations; the executeQuery tool can run INSERT/UPDATE/DELETE statements on tables that support them, so the LLM may modify connected data. The HTTP/SSE transport binds to 127.0.0.1:8070 by default and supports a --no-auth flag and an internet tunnel (--tunnel); exposing the server beyond localhost without authentication grants query access to all connected data.
Anyquery FAQ
How do I start the MCP server?
Run `anyquery mcp --stdio` (started by your MCP client) for stdio transport, or `anyquery mcp --host 127.0.0.1 --port 8070` for an HTTP/SSE tunnel.
Which clients can connect?
Any MCP-compatible client such as Claude Desktop, Cursor, or Windsurf. Anyquery also supports function-calling clients like ChatGPT and TypingMind via `anyquery gpt`.
How do I add new data sources?
Install plugins from the official Anyquery registry (40+ integrations) or load any SQLite extension. You can also pass a SQLite database with the `--database` flag.
Can the LLM modify my data?
Yes — the executeQuery tool supports INSERT/UPDATE/DELETE on tables that allow those operations, so the model can write to connected data, not just read it.
Alternatives to Anyquery
Compare all alternatives →
Google's official MCP server with prebuilt BigQuery tools, querying datasets via Application Default Credentials.
Create, manage, and query your InstantDB apps, schemas, permissions, and data from your AI editor.
Compare Anyquery with: