observability

Observability MCP Server provides a unified MCP interface to monitor and manage observability-related systems such as Prometheus, Loki, Nacos, MySQL, Doris, Kubernetes, and more. Built in Rust for performance and reliability, it exposes HTTP endpoints (including SSE) to query metrics, fetch logs, and provision configuration across supported data sources. After building, run the released binary and point clients to the MCP endpoints exposed by the server. The README notes a default web access path at http://localhost:3013/mcp, where you can discover available MCP tools and endpoints. The server aims to offer a single, coherent API surface for querying metrics, logs, and metadata across multiple systems, with a focus on real-time data and documentation integration.

Key capabilities include Prometheus querying (query_instant, query_range, series_metadata, label_values, alertmanagers), documentation tooling (docs_list, docs_read, docs_search), and ongoing integration work for additional backends like Loki, Nacos, MySQL, Doris, Kubernetes, RocketMQ, and Kafka. Use the provided MCP Tools section in the README as a reference for the specific endpoints you can call via MCP. Tools are designed to be invoked through HTTP requests and can be integrated into dashboards, alerting pipelines, or automation scripts.

Prerequisites:

• Rust toolchain installed (Rust compiler and Cargo) for building from source.
• Basic familiarity with building Rust projects and running binaries.

Installation steps:

1. Clone the repository: git clone <repository-url> cd observability-mcp-server

2. Build the project in release mode: cargo build --release

3. Run the binary from the target directory: ./target/release/observability-mcp-server

4. Verify the server is running by accessing: http://localhost:3013/mcp

Environment configuration (optional):

• You can configure connections to backend systems (Prometheus, Loki, Nacos, etc.) via environment variables as documented in the README. Typically, you would set variables like PROMETHEUS_ROOT, LOKI_ROOT, and others to point the MCP server to the appropriate endpoints before starting the server.

Notes:

• The README mentions some integrations are marked as coming soon (e.g., Loki, RocketMQ, Kafka), so some endpoints may be placeholders until those backends are fully implemented.

Tips and considerations:

• The server is designed to run with a single HTTP transport and SSE support, enabling real-time streaming of data where supported by the backend.
• If Prometheus integration is critical, ensure the PROMETHEUS_ROOT variable is correctly set to the Prometheus HTTP API endpoint.
• For any backend that is marked as coming soon (Loki, RocketMQ, Kafka, etc.), monitor the project for updates and enable/disable features accordingly via environment configuration.
• Since the Quick Start shows a manual build and run flow, consider wrapping the run command in a systemd service or supervisor to keep the MCP server running in production.
• Check the project structure under src for where MCP tools are defined and how to extend or customize the available endpoints for new platforms.