How zoro-zuro/incident-a works
This is a local-first incident response agent: alerts enter through Redis or the REST API, then an orchestrated workflow investigates, recommends a response, waits for operator approval, and records the outcome for future recall. The workflow includes severity classification, historical-incident recall, evidence gathering, runbook retrieval, decision-making, remediation, verification, and reporting. The API exposes incident and knowledge views to a React dashboard. The README documents the end-to-end flow; sampled source confirms several calls and integrations, but not every workflow transition or API implementation detail.
Main entry points: graph.py Workflow Orchestrator · main.py REST API · memory.py Incident Memory · incident_runner.py Incident Runner
Components 23
Alert Ingestion
- Redis Alerts
- Queue Helpers
agent/redis_queue.py - Alert Listener
agent/listener.py
Incident Workflow
- Incident Runner
agent/incident_runner.py - Workflow Orchestrator
agent/graph.py - Incident State
agent/state.py
Knowledge & Evidence
- Incident Memory
agent/memory.py - Runbook Retrieval
agent/rag.py - Embedding Helper
agent/embeddings.py - Evidence Tools
agent/tools.py - Metrics Queries
agent/metrics.py - Simulated Services
agent/simulated_env.py
Decision & Response
- Decision Provider
agent/decision_provider.py - LLM Adapter
agent/llm.py - Reasoning & Retry
agent/reasoner.py - Human Approval
agent/approval.py - Remediation
agent/remediation.py
API & Dashboard
- REST API
api/main.py - React Dashboard
frontend/src/App.tsx - API Client
frontend/src/lib/api.ts - Incident Views
frontend/src/components/IncidentDetails.tsx
Other components
- Alert Source
- ChromaDB
Connections 29
- Alert Source → Redis Alerts: publishes alerts
README.md - Redis Alerts → Queue Helpers: delivers alerts
README.md - Queue Helpers → Alert Listener: dispatches alerts
agent/listener.py - Alert Listener → Incident Runner: starts workflow
agent/listener.py - Alert Source → REST API: submits alert
README.md - REST API → Incident Runner: creates incident
api/main.py - Incident Runner → Workflow Orchestrator: executes workflow
agent/incident_runner.py - Incident Runner → Incident State: tracks state
agent/incident_runner.py - Workflow Orchestrator → Incident Memory: recalls incidents
agent/graph.py - Workflow Orchestrator → Runbook Retrieval: retrieves runbook
agent/graph.py - Workflow Orchestrator → Evidence Tools: gathers evidence
README.md - Workflow Orchestrator → Decision Provider: evaluates incident
agent/graph.py - Workflow Orchestrator → Human Approval: requests approval
agent/graph.py - Workflow Orchestrator → Remediation: runs remediation
README.md - Workflow Orchestrator → Incident Memory: stores outcome
README.md - Incident Memory → ChromaDB: reads and writes
agent/memory.py - Runbook Retrieval → ChromaDB: searches vectors
agent/rag.py - Runbook Retrieval → Embedding Helper: uses embeddings
agent/rag.py - Incident Memory → Embedding Helper: uses embeddings
agent/memory.py - Evidence Tools → Simulated Services: queries services
agent/tools.py - Metrics Queries → Simulated Services: queries metrics
agent/metrics.py - Remediation → Simulated Services: changes services
agent/remediation.py - Decision Provider → LLM Adapter: uses model adapter
- Workflow Orchestrator → Reasoning & Retry: uses retry logic
agent/graph.py - React Dashboard → API Client: loads data
frontend/src/App.tsx - Incident Views → API Client: fetches incident
frontend/src/components/IncidentDetails.tsx - API Client → REST API: calls endpoints
frontend/src/lib/api.ts - REST API → Incident Memory: queries memory
api/main.py - REST API → Queue Helpers: publishes alerts
api/main.py