Alibabacloud Observability Mcp Server

What is Alibabacloud Observability Mcp Server?

It is a Go implementation of the Alibaba Cloud Observability MCP Server that provides AI models with structured data access to Alibaba Cloud Log Service (SLS) and Cloud Monitor (CMS). Based on the Model Context Protocol, it integrates seamlessly with AI tools such as Cursor, Kiro, Cline, and Windsurf.

How to use Alibabacloud Observability Mcp Server?

Download the binary from the Releases page, set the ALIBABA_CLOUD_ACCESS_KEY_ID and ALIBABA_CLOUD_ACCESS_KEY_SECRET environment variables, then start the server in stdio mode (./alibabacloud-observability-mcp-server start --stdio) or using a config file (./alibabacloud-observability-mcp-server start --config config.yaml). Alternatively, build from source and configure tool integration via mcp.json.

Key features of Alibabacloud Observability Mcp Server

• Supports stdio, SSE, and streamable-http transport modes.
• Modular toolset: PaaS, IaaS, Shared.
• Flexible time expression parsing for data queries.
• Time series comparison with statistical analysis and trend scoring.
• Structured JSON logging (slog) and graceful shutdown.
• Single binary with zero runtime dependencies.

Use cases of Alibabacloud Observability Mcp Server

• Natural language query over SLS logs and CMS metrics via AI agents.
• Automated root cause analysis by translating user prompts into SQL/SPL.
• Time series data comparison and anomaly detection in observability dashboards.
• Integration with coding assistants for intelligent operations and troubleshooting.

FAQ from Alibabacloud Observability Mcp Server

What credentials are required?

You need an Alibaba Cloud AccessKey ID and Secret, set via the ALIBABA_CLOUD_ACCESS_KEY_ID and ALIBABA_CLOUD_ACCESS_KEY_SECRET environment variables.

Which tools are paid and how are they billed?

The AI-powered tools (sls_text_to_sql, sls_text_to_spl, sls_sop, cms_natural_language_query) incur STAROps fees per call. Free tools can be selected by configuring enabled_tools in config.yaml.

What time expression formats are supported?

Supported formats include relative presets (last_5m), relative expressions (now()-1h), Grafana style (now-15m~now-5m), absolute timestamps (seconds/milliseconds), and date-time strings (2024-01-01 00:00:00).

Is the server secure and does it store credentials?

The server does not store AccessKeys; they are used only at runtime for API calls. It is recommended to deploy inside a VPC or intranet and apply access control when using SSE/HTTP modes.

What transport modes can I use to integrate with AI tools?

The server supports stdio, SSE, and streamable-http transports. For streamable-http (recommended), set server.transport: streamable-http in config.yaml. For stdio, use the command-line invocation with the --stdio flag.