Sumo Logic MCP. Run deep log searches and manage infra from chat.

Your AI client connects your agent directly to your log and metric data stream through Sumo Logic. This server lets you manage everything from deep log searches to core infrastructure tasks—all using plain chat commands. You don't gotta jump between dashboards just to find an error code or check a system status.

The agent handles it all.

Deep Log Searches
When you need to dig into huge datasets, the agent runs background jobs for deep log searches by calling create_search_job. It doesn’t give you results right away; that's how these things work. To see what's going on with that job, you check its status using get_search_job_status until it says it's done.

Once the search is complete, you can pull the raw logs individually with get_search_job_messages, or you can get a summary record count by running get_search_job_records. If that job runs too long and costs you dough, you gotta hit delete_search_job to stop it.

System Monitoring & Metrics
If you're checking system health, the agent executes metrics queries using execute_metrics_query. You feed it a query, and it spits out time-series data about how your system is performing. That’s where you monitor performance without writing complex dashboard filters.

Infrastructure Control: Collectors and Sources
Managing your log ingestion points—your collectors and sources—is straightforward. To see what's connected, the agent lists all installed or hosted collectors by running list_collectors. You can grab detailed info on a specific collector using get_collector for its ID, or you can set up a whole new data endpoint in the cloud environment with create_hosted_collector.

If something changes with an existing setup, you'll use update_collector; and if that entire endpoint is trash, you delete it permanently with delete_collector.

For the specific streams inside those collectors, the agent manages sources. You can view every active data stream tied to a collector using list_sources. To get details on one source, use get_source. If you gotta change the parameters of an existing log feed, run update_source. Need to add a whole new data feed? Use create_source.

And if that specific source is done, delete it with delete_source.

User and Access Management
The agent keeps your organization secure by handling user accounts. You can see everyone currently set up using list_users. If you need to bring someone on board, use create_user. When an employee leaves or loses access, you can deactivate their account with delete_user.

It's simple: instead of running commands in a console and getting lost in the documentation, you just tell your AI client what job needs doing. It runs the command—whether it’s listing all collectors, checking a metric query, or pulling raw log messages—and hands you the answer. You get full observability without leaving plain chat.