The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions specify that the agent should ingest and analyze untrusted external data, such as cyber threat intelligence (CTI) reports, STIX files, and detection logs, to map them to the MITRE ATT&CK framework.\n
Ingestion points: Instructions in SKILL.md direct the agent to read supplied information, reports, appendixes, and linked STIX files.\n
Boundary markers: The instructions lack explicit requirements for the agent to use delimiters or to treat embedded instructions in the processed reports as data rather than instructions.\n
Capability inventory: The skill primarily uses grep for searching its own local resource files. No network operations, arbitrary file-writing, or system-altering capabilities are identified in the provided instructions or resource files.\n
Sanitization: There are no specified procedures for sanitizing, escaping, or filtering the content of ingested external reports before analysis.

analysing-attack