What problem does it solve?
Answering aggregate questions across a packet capture — top talkers, service counts, alert frequencies — is impractical when reading Zeek logs or Suricata eve.json line by line. This Skill loads exported network security data into a sandboxed DuckDB engine so you can run SQL aggregations, joins, and counts across an entire capture.
Core Features & Use Cases
- SQL over Zeek logs: Query conn, http, and other Zeek tables with their real column names, joining on
uid to pivot across sessions.
- Suricata alert aggregation: Group and count IDS alert signatures from eve.json to see which detections fired most often.
- Locked read-only sandbox: DuckDB runs with external access disabled and configuration locked, so queries can never reach memory stores, credentials, or arbitrary files.
- Use Case: After running a packet capture analysis, ask which hosts generated the most connections and which Suricata signatures fired, then join HTTP requests to connection metadata to reconstruct suspicious sessions.
Quick Start
Ask the agent to show the top 20 talkers by connection count from the latest Zeek capture run.