The Agent Skills Directory

[DATA_EXFILTRATION]: The skill performs legitimate network operations to established security platforms (VirusTotal, AbuseIPDB, MalwareBazaar) to enrich artifact data. These operations are core to the skill's purpose and use standard industry APIs.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests and processes untrusted IOC data from external sources or user input.
Ingestion points: IOC strings provided in the demo list and potentially supplied via user input at runtime.
Boundary markers: The skill does not currently implement specific delimiters to wrap untrusted content when presenting enrichment results to the agent.
Capability inventory: The skill has the capability to perform network requests to various threat intelligence APIs using the requests library.
Sanitization: Implements specific defanging logic (e.g., replacing dots with [.] and protocol headers with hxxp) to ensure malicious strings are safely documented and not accidentally clicked or interpreted as active links.
[EXTERNAL_DOWNLOADS]: The skill references several standard, well-known Python packages including requests, vt-py, and pymisp for its threat intelligence operations.
[SAFE]: The malicious URL identified in automated scans (http://malicious-site.com/payload.exe) is utilized as a static test indicator for classification and enrichment exercises; it is not directly accessed or executed by the skill code.
[SAFE]: A minor metadata discrepancy exists between the author name in the manifest (mahipal) and the license/context (mukul975), which does not pose a security risk.

analyzing-indicators-of-compromise