The Agent Skills Directory

[SAFE]: The skill implements its core functionality using well-documented libraries and official threat intelligence data sources. No patterns indicative of command injection, credential theft, or unauthorized network access were found.
[INDIRECT_PROMPT_INJECTION]: The script reads external data from MITRE ATT&CK servers. 1. Ingestion points: Technique and group descriptions are retrieved in scripts/agent.py via the attackcti library. 2. Boundary markers: No delimiters are used to wrap the fetched technique descriptions. 3. Capability inventory: The skill writes JSON files to the local filesystem and prints data to the console. 4. Sanitization: Descriptions are truncated for display but otherwise processed as raw strings. The overall risk is considered safe due to the trusted nature of the official MITRE data source.

analyzing-threat-actor-ttps-with-mitre-navigator