suse-observability

This SUSE Observability MCP Server provides AI agents with structured access to SUSE Observability (StackState) data. It exposes tools to inspect topology, query real-time and historical metrics, and monitor health states. Agents can discover components, fetch related metrics, pull monitor states, and perform root-cause analysis by combining topology with time-series data. The available tools are designed to be used by MCP clients to gather context, run PromQL-like queries, and present results in a model-context friendly format for downstream reasoning.

Prerequisites:

• Docker (recommended for quick Start) or Go toolchain if you prefer building from source
• Access to a SUSE Observability instance with a valid API token

Option A: Run via Docker (recommended)

1. Build or pull the image (if building locally, use the provided Dockerfile): docker build -t suse-observability-mcp-server .
2. Run the container (adjust URL/token as needed): docker run -p 8080:8080 --rm suse-observability-mcp-server
   -url "https://your-instance.suse.observability.com"
   -token "YOUR_API_TOKEN"
   -apitoken
   -http ":8080"

Option B: Build from source (Go)

1. Install Go 1.23+ from the official site.
2. Clone the repository: git clone <repository-url> && cd <repo-directory>
3. Build the server: go build -o suse-observability-mcp-server cmd/server/main.go
4. Run the binary (stdio by default or enable HTTP): ./suse-observability-mcp-server
   -url "https://your-instance.suse.observability.com"
   -token "YOUR_API_TOKEN"
   -apitoken

Configuration flags to be aware of:

• -http: Bind address for HTTP transport (e.g., ":8080"). If omitted, stdio mode is used.
• -url: SUSE Observability API URL
• -token: API token for SUSE Observability
• -apitoken: Use API token instead of a Service Token (boolean)

Notes and tips:

• Ensure the API token has the required scopes for topology, metrics, and monitors data access.
• When running in Docker, exposing the correct port (e.g., 8080) is important if you plan to use HTTP transport.
• The server normalizes technical identifiers into LLM-friendly formats to reduce hallucinations; verify results against the source SUSE Observability data when performing critical root-cause analysis.
• If you encounter authentication failures, double-check token type and that the token is valid for the target SUSE Observability instance.
• For large queries, consider adjusting the step or time range in getMetrics to manage pagination and response size.