Introduction

auralogs is the logger built for AI agents. Drop the SDK into your app, send structured logs to auralogs, and expose them through an MCP server so Codex, Claude, Cursor, and other agents can investigate production issues during a conversation.

What you get

• A lightweight SDK — one import, one init call. Captures logs, errors, and unhandled exceptions automatically.
• MCP access for agents — connect a read-only MCP client and ask it to search logs, inspect metadata, pull related events, and summarize what happened.
• Dashboard and REST access — use the log viewer for humans, or the REST API for scripts, dashboards, and incident tooling.
• Optional AI analysis — add an Anthropic or OpenAI key when you want auralogs to generate analyses, alerts, and scheduled reviews.
• Optional GitHub autofix — connect a repo later if you want agents to trace incidents back to source and open fix PRs.

How it works

1. Install the SDK — npm install auralog-sdk and add two lines to your entry point.
2. Logs flow to auralogs — the SDK batches and sends logs to the ingest endpoint. Errors are sent immediately.
3. Copy a read key — new projects show an MCP/API read key once; read keys are scoped to one project and can only read logs, analyses, and project metadata.
4. Connect your agent through MCP — point Codex, Claude, Cursor, or another MCP client at https://mcp.auralog.ai/mcp and ask questions about production behavior.
5. Add automation when needed — enable AI analysis, notifications, webhooks, or GitHub autofix after the logging and MCP loop is working.

Get started

Head to the Quick Start to have auralogs running in under 2 minutes.