Let your AI agent debug production logs with read-only MCP.
Connect Codex, Claude, Cursor, or an internal agent to Auralogs. It can search production logs, inspect metadata, and explain failures without copy-paste. Read keys are scoped to one project and do not grant source or write access.
Give agents read-only access to logs, analyses, and project metadata.
Ask natural-language debugging questions instead of pasting stack traces.
Keep GitHub and source access optional until you want autofix PRs.
How it works
Start with structured logging, then choose how much agent automation you want on top of it.
Install the SDK
Send structured logs, errors, trace IDs, and user metadata from your app to Auralogs.
Create a read key
Generate an MCP/API read key scoped to one project. It can query context, not mutate settings.
Connect your agent
Add the Auralogs MCP endpoint to Codex, Claude, Cursor, or any MCP client with Streamable HTTP support.
Questions you can answer
Auralogs keeps the raw logs, metadata, and analysis context close enough for humans and agents to use.
Why did checkout errors spike after the last deploy?
Find error logs for trace_9f3 and summarize the pattern.
Compare staging and production failures for this endpoint.
FAQ
Which AI agents can use Auralogs?
Auralogs works with Codex, Claude, Cursor, and other MCP clients that support Streamable HTTP with bearer-token authentication.
Does the agent need GitHub access?
No. MCP investigation only needs a read key for logs and analyses. GitHub access is optional and only needed for autofix PRs.
Can agents see every project?
No. Read keys are scoped to one project, so you can create separate keys for different agents, workspaces, or environments.
Try it with your next error spike
Create a free project, send your first logs, and copy the MCP/API read key for your agent.
Start free