What problem does it solve?
Security teams need continuous detection and investigation across security log data, but manual triage and analyst workflows cannot keep up with fast-changing threats and noisy alerts.
Core Features & Use Cases
- RAG-based behavioral memory: Builds and queries embeddings of normal network behavior stored in OpenSearch/Elasticsearch for contextual detection.
- LLM anomaly analysis: Uses LLM validation to analyze anomalies with retrieved context and produce investigation-ready outputs.
- LangGraph orchestrated SOC automation: Runs a DECIDE→EXECUTE→EVALUATE supervisor loop with conversation-based investigations and checkpointing.
- Operational interfaces: Provides both CLI and a web UI/API to support real-time threat analysis and automation.
- Provider flexibility: Supports OpenSearch vs Elasticsearch and Ollama vs other LLM providers.
Example use case
Monitor OpenSearch indices for near-real-time network anomalies, retrieve baseline context for the involved events, and automatically generate an analyst-style threat triage and investigation plan.
Quick Start
Deploy the SecurityClaw service by running the command: python main.py service.