How devlup-labs/inframind works
InfraMind is an autonomous incident-response workflow for a mock ML service on Kubernetes. It monitors Prometheus metrics, gathers pod logs when it detects an anomaly, uses retrieved log evidence to identify a likely cause and affected pod, then decides whether a constrained mitigation is safe. The loop rechecks metrics and produces a human-readable incident report. The README documents these capabilities and relationships; no source excerpts were available, so component paths point only to the visible agents directory.
Main entry points: agents Optimization Agent
Components 13
Telemetry
- Mock ML service
- Prometheus metrics
- Monitoring Agent
agents
Investigation
- Log Collector
agents - Qdrant retrieval
- Root Cause Agent
agents
Remediation
- Optimization Agent
agents - Horizontal scaling
agents - Pod restart
agents - Kubernetes
- No safe action
agents
Reporting
- Reporting Agent
agents - Incident report
Connections 16
- Mock ML service → Prometheus metrics: exposes metrics
README.md - Monitoring Agent → Prometheus metrics: reads metrics
README.md - Monitoring Agent → Log Collector: triggers on anomaly
README.md - Log Collector → Kubernetes: collects pod logs
README.md - Log Collector → Qdrant retrieval: stores log embeddings
README.md - Root Cause Agent → Qdrant retrieval: searches log evidence
README.md - Root Cause Agent → Optimization Agent: provides cause and pod
README.md - Optimization Agent → Horizontal scaling: selects for pressure
README.md - Optimization Agent → Pod restart: selects for pod failure
README.md - Optimization Agent → No safe action: declines unsafe action
README.md - Horizontal scaling → Kubernetes: scales pods
README.md - Pod restart → Kubernetes: restarts affected pod
README.md - No safe action → Reporting Agent: continues incident loop
README.md - Horizontal scaling → Monitoring Agent: waits and rechecks
README.md - Pod restart → Monitoring Agent: waits and rechecks
README.md - Reporting Agent → Incident report: produces report
README.md