MCP Server

What is MCP Server?

An MCP server for Chronosphere that provides tools for fetching logs, metrics, traces, events, and select entities. It is a work in progress – many features are not yet implemented and may change without warning.

How to use MCP Server?

Build the binary with make chronomcp, then configure it as an MCP server in your client (e.g., Claude Desktop or Cursor) by providing the path to the binary, a config file, and environment variables CHRONOSPHERE_ORG_NAME and CHRONOSPHERE_API_TOKEN. Alternatively, run directly with make run-chronomcp.

Key features of MCP Server

• Fetches logs, metrics, traces, and events from Chronosphere
• Also retrieves select Chronosphere entities
• Uses standard Chronosphere authentication (API token)
• Supports Streamable HTTP transport for debugging
• Includes experimental Librechat and agent integrations

Use cases of MCP Server

• Querying Chronosphere observability data through an MCP client
• Embedding Chronosphere data retrieval into AI‑powered chat interfaces (experimental)
• Debugging and inspecting MCP tool calls via the MCP Inspector

FAQ from MCP Server

What dependencies or runtime does MCP Server require?

It requires Go to build (make chronomcp). The resulting binary runs as a standalone process with environment variables for organization name and API token.

How does MCP Server authenticate to Chronosphere?

It uses a Chronosphere API token, provided via the CHRONOSPHERE_API_TOKEN environment variable. The same token file used by chronoctl is supported.

What transports and authentication does MCP Server support?

The README shows use with a local binary via stdio (Claude Desktop, Cursor) and streamable HTTP for debugging. Authentication is handled by the API token set in environment variables.

What are the known limits of MCP Server?

The project is labeled as a work in progress. Many features are not yet implemented, and existing ones may be added, changed, or removed without warning.

Where does MCP Server data live?

Data is fetched from the user’s Chronosphere organization at query time; no data is stored locally by the server aside from temporary results.