The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to perform deep analysis on untrusted external artifacts, creating a potential vector for indirect prompt injection through malicious payloads in analyzed source code or binaries. * Ingestion points: The skill reads repository files, binary metadata, and network service responses during 'Phase 1: Target Profiling' and 'Phase 3: Deep Analysis'. * Boundary markers: There are no explicit markers or instructions to isolate the agent from potentially malicious instructions embedded in the targets being researched. * Capability inventory: The skill employs powerful analysis tools such as grep, find, nmap, and various binary analysis utilities. * Sanitization: The skill lacks sanitization mechanisms for the untrusted data it processes via shell commands.
[EXTERNAL_DOWNLOADS]: The skill references external resources for tool installation and vulnerability intelligence. * Evidence: references/binary-analysis.md includes documentation for installing the GEF extension for GDB via a shell script from its official domain. * Evidence: SKILL.md fetches vulnerability advisories from the NIST National Vulnerability Database, a well-known and trusted service.

vulnerability-research