Skip to main content

Resources

The Alertmanager MCP Server provides 11 MCP Resources for real-time data streams. Resources are read-only — they provide context for AI agents to make informed decisions before using state-changing tools.

Available Resources​

Resource URIDescription
am://system/backendsAll known backends with health status — start here
am://system/backends/{backend_id}Detailed status, version, cluster info for one backend
am://system/statusAlertmanager version, uptime, cluster info, config summary
am://system/receiversConfigured receivers (Slack, PagerDuty, email, webhook) with redacted config
am://system/configRouting tree and inhibition rules (secrets redacted)
am://system/audit-logRecent MCP-initiated operations
am://alerts/activeBounded snapshot of active alerts
am://alerts/groupsSnapshot of alert groups as computed by Alertmanager
am://silences/activeSnapshot of active silences
am://best-practicesAlerting best practices reference
am://onboarding-guideAlert onboarding guide

am://system/backends​

All configured Alertmanager backends with health status. Always read this first to discover available backend_id values.

Use Case:

"What Alertmanager backends are available?"
"Is our production Alertmanager healthy?"

am://system/status​

Alertmanager version, uptime, cluster peer status, and configuration summary.

Use Case:

"What version of Alertmanager is running?"
"How many cluster peers are in the production Alertmanager?"

am://system/receivers​

Lists all configured receivers with their integration types (Slack, PagerDuty, email, webhook). Configuration values are redacted for security.

Use Case:

"What Slack channels are configured for alerts?"
"List all PagerDuty receivers."
"What notification integrations are set up?"

am://system/config​

Full routing tree and inhibition rules. Secrets and sensitive values are redacted.

Use Case:

"Show me the full routing tree."
"What inhibition rules are configured?"
"Export the current Alertmanager config for Git storage."

am://system/audit-log​

In-memory log of all operations initiated through this MCP server — silence creation, expiration, extension, and test alert injection.

Use Case:

"What MCP operations have been performed recently?"
"Show the audit log for this session."

am://alerts/active​

Bounded snapshot of currently active (firing) alerts. Faster than a tool call for quick status checks.

Use Case:

"What alerts are currently firing?"
"Give me a quick snapshot of active alerts."

am://alerts/groups​

Alert groups as computed by Alertmanager — respecting group_by labels from the routing configuration.

Use Case:

"How are alerts grouped right now?"
"Show me the alert group structure."

am://silences/active​

Snapshot of all currently active silences with matchers, expiry times, and authors.

Use Case:

"Are there any active silences right now?"
"Who created the current silences?"

Using Resources​

Resources are accessed via the MCP protocol using their URI. AI agents read resources for context before making tool calls.

Client: What's firing right now?
Agent: [Reads am://alerts/active]
       [Returns active alert snapshot]
Client: Who gets paged for checkout alerts?
Agent: [Reads am://system/receivers]
       [Then calls am_explain_routing with checkout labels]

Next Steps​