A walkthrough shows how AppSignal's MCP server lets AI agents like Claude Code or Cursor query production monitoring data directly, turning a typical 10-15 minute manual incident investigation into a sub-3-minute conversational workflow. The example traces a payments service error spike back to a Stripe API latency issue, with the agent pulling error rates, backtraces, deploy markers, and metrics, then documenting and assigning the incident. A setup guide, prompt cheat sheet, capability limits, best-practice tips, and a comparison against Scout's and Honeybadger's MCP servers round out the piece, noting AppSignal is the only one of the three that lets an agent both query and act on incidents.
Table of contents
MCP and Its Role in Incident ResponseSetup in Two MinutesReal Incident InvestigationPrompt Cheat SheetWhat AppSignal MCP Can and Cannot DoTips for Getting the Most Out of MCPNext StepsFrequently Asked Questions (FAQ)Questions this post answers
How does AppSignal's MCP server compare to Scout's and Honeybadger's MCP servers?
AppSignal's MCP server is the only one of the three that lets an agent both query and act on incidents. Scout's MCP server is read-only across metrics, traces, jobs, errors, and anomalies with no way to update incidents. Honeybadger's is scoped to error and fault data only, with write access limited to project administration rather than managing faults or incidents. Comparing monitoring tools with agent support? daily.dev surfaces this kind of head-to-head breakdown.
What tools can an AI agent call through AppSignal's MCP server during an incident?
An agent connected via AppSignal's MCP server can call tools like get_app_resources for deploy markers, get_metrics_timeseries for performance data, and manage_incident_note or update_incidents to document and assign incidents. It can query errors, stack traces, performance data, host metrics, and deploy markers, then correlate them using natural language filters. Developers wiring monitoring into their AI agents can track workflows like this on daily.dev.
Can an AI agent connected via MCP trigger a rollback or restart a service during a production incident?
No, an AI agent using AppSignal's MCP server cannot trigger rollbacks, restart services, or take action outside AppSignal unless a separate tool is connected and chained to it. Write access through this integration is limited to AppSignal's own data, such as incidents, anomaly triggers, log actions, and dashboards, and cannot touch infrastructure directly. daily.dev helps engineers weighing what AI agents should and shouldn't be trusted to act on in production.