Auralogs

What Auralogs MCP server does

Auralogs MCP server gives AI agents read-only access to data held in the Auralogs logging platform. It is intended for production investigation: an agent can find recent errors, search messages for a pattern, inspect stack traces and metadata, and review analyses generated for operational issues.

The available operations cover both project discovery and incident inspection. Agents can list projects available to the supplied read key or retrieve a project by ID. They can also list logs by level and time range, search log messages with full-text search, and fetch an individual log with its surrounding context. Analysis tools list results by severity or type and retrieve a single analysis containing its diagnosis, suggested fix, and PR link when available.

How it works

Auralogs MCP server is hosted by Auralogs and uses Streamable HTTP rather than a locally installed process. Clients connect to https://mcp.auralogs.ai/mcp and send an Authorization header containing a bearer token. The token is a read key scoped to one project, so the accessible data depends on the key used for the connection.

The MCP surface contains no write operations. Agents can inspect information but cannot modify logs, projects, or analyses through this server. The documented limit is 60 requests per minute for each key. A revoked key returns 401 on the next request.

Setup and configuration

Create an Auralogs account, create a project, and copy the MCP/API read key from the startup dialog or from Settings → API & MCP keys → Read keys. Keys use the aura_read_... format and are shown in plaintext only when created. Add the remote MCP URL and the key as an HTTP authorization header in the client configuration.

The README provides connection examples for Claude Code, Claude Desktop, Cursor, and Cline. Codex users can use the separate Auralogs Codex plugin or a standard remote MCP configuration. A basic curl initialization request can verify the endpoint, authentication, and transport; a successful response includes server information with the name auralogs.

Tools and capabilities

• list_projects: identifies projects accessible through the current read key.
• get_project: retrieves a project by ID.
• list_logs: returns logs filtered by severity level and time range.
• search_logs: searches log messages using full-text matching.
• get_log: retrieves one log with stack and metadata context.
• list_analyses: filters AI analyses by severity and analysis type.
• get_analysis: retrieves an analysis with its diagnosis, suggested fix, and PR link.

Limitations and notes

Auralogs MCP server only exposes read operations; it cannot apply fixes or change Auralogs data. Each read key is limited to one project, and access is controlled by that key. The server is hosted, so the repository contains metadata and connection instructions rather than a local server implementation. Requests are subject to the 60-per-minute, per-key rate limit.