implementing-network-traffic-analysis-with-arkime

Warn

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The agent script explicitly disables SSL certificate verification when communicating with the Arkime API. This allows for Man-in-the-Middle (MitM) attacks where an attacker could intercept the Digest authentication credentials or the network traffic data being analyzed.
  • Evidence: In scripts/agent.py, the arkime_request function sets verify=False in the requests.get call.
  • [COMMAND_EXECUTION]: The skill design requires passing sensitive credentials such as passwords as plaintext command-line arguments. This is insecure as CLI arguments are often visible in system process lists and stored in shell history files.
  • Evidence: Both SKILL.md (example command) and scripts/agent.py (argparse setup) use the --password flag for authentication.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes data from an external network source (Arkime API) without sanitization.
  • Ingestion points: API data fetched in scripts/agent.py from /api/sessions, /api/connections, and /api/spiview.
  • Boundary markers: None.
  • Capability inventory: The script writes analysis results to a local file using json.dump.
  • Sanitization: None; data from the API is directly formatted into the report without validation or filtering.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 16, 2026, 03:46 PM