Connecting ChatGPT/Claude
LLMs work best when given enough context to generate meaningful responses. Anyquery is a query engine for all your applications. By connecting it to ChatGPT/Claude, you can interact with your applications using natural language.
For example, you can let ChatGPT to access your GitHub repositories.
Prerequisites
Section titled “Prerequisites”Before you begin, ensure that you have the following:
- A working installation of Anyquery (
v0.4.0). Follow the installation guide if you haven’t installed it yet. - A few integrations set up in Anyquery. Refer to each integration guide for more information.
Connecting your chat interface to Anyquery
Section titled “Connecting your chat interface to Anyquery”Anyquery supports function calling and the Model Context Protocol (MCP) to connect to your chat interface. Refer to each connection guide in the documentation for more information.
MCP transports
Section titled “MCP transports”anyquery mcp can talk to your client in three ways:
| Transport | How to start it | Address |
|---|---|---|
| Standard input/output | anyquery mcp --stdio |
Launched by the client |
| Streamable HTTP (recommended for HTTP) | anyquery mcp |
http://127.0.0.1:8070/mcp |
| Server-sent events (legacy) | anyquery mcp |
http://127.0.0.1:8070/sse |
Both HTTP endpoints are served on the same port, so older clients that only support SSE keep working. Change the address with the --host and --port flags. HTTP connections require the bearer token printed at startup (or set with the ANYQUERY_AI_SERVER_BEARER_TOKEN environment variable) in the Authorization header, unless you pass --no-auth.
Anyquery supports the MCP protocol versions 2026-07-28, 2025-11-25, 2025-06-18, 2025-03-26 and 2024-11-05, and negotiates the version with each client.
Once connected, you can interact with your applications using natural language. For example, if you’ve set up the Spotify integration, you can ask ChatGPT to recommend a song for you.
Here are a few examples:
- [Spotify] Recommend me a song according to my history.
- [GitHub] Which issues are assigned to me?
- [Reminders] Remind me to call John at 3 PM.
- [Calendar] What’s on my schedule for tomorrow? Will I be free at 2 PM?
- [Browser] Take this PDF, and open the most important links in my browser.
- [Email] Can you summarize the email from John?
- [Files] Find the document I worked on last week. It’s a PDF file with the title “Project Plan”.
- [Notion] Retrieve Franck’s notes in my CRM database. I have a meeting with him tomorrow.
- [Google Sheets] Update the sales report with the latest data.