The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled authentication logs to detect behavioral anomalies, which could theoretically be influenced by an attacker to trigger or evade alerts.\n
Ingestion points: scripts/agent.py ingests data from Elasticsearch logs.\n
Boundary markers: None present; the script performs statistical analysis on raw field values like geolocations and timestamps.\n
Capability inventory: The script calculates risk scores and prints alerts to the console.\n
Sanitization: Not present for the input log data, as the script expects pre-normalized SIEM data.\n- [SAFE]: The Python script scripts/agent.py contains a minor best-practice violation by disabling SSL certificate verification for its data source connection.\n
Evidence: The get_es_client function initializes the Elasticsearch client with verify_certs=False.\n
This configuration is common in internal environments using self-signed certificates and does not indicate malicious intent.

performing-user-behavior-analytics