groups:
- name: thoughtbox-mcp
rules:
- alert: ThoughtboxHighErrorRate
expr: (sum(rate(mcp_requests_total{status="error", server_name="thoughtbox"}[5m])) / sum(rate(mcp_requests_total{server_name="thoughtbox"}[5m]))) > 0.05 and sum(rate(mcp_requests_total{server_name="thoughtbox"}[5m])) > 0
for: 5m
labels:
severity: warning
annotations:
summary: "Thoughtbox error rate > 5%"
description: "Error rate is {{ $value | humanizePercentage }}"
- alert: ThoughtboxLatencyP95High
expr: histogram_quantile(0.95, sum by (le) (rate(mcp_request_duration_seconds_bucket{server_name="thoughtbox"}[10m]))) > 5
for: 10m
labels:
severity: warning
annotations:
summary: "Thoughtbox p95 latency > 5s"
description: "P95 latency is {{ $value }}s"
- alert: ThoughtboxDown
expr: mcp_upstream_available{server_name="thoughtbox"} == 0
for: 2m
labels:
severity: critical
annotations:
summary: "Thoughtbox MCP server is DOWN"
description: "Upstream thoughtbox server is not responding"
- alert: ThoughtboxToolCallSlow
expr: histogram_quantile(0.95, sum by (le, tool_name) (rate(mcp_tool_duration_seconds_bucket{server_name="thoughtbox"}[10m]))) > 10
for: 5m
labels:
severity: warning
annotations:
summary: "Tool {{ $labels.tool_name }} p95 > 10s"
description: "Tool call latency is {{ $value }}s"
