The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze data from external, untrusted sources including CVE databases (NVD, OSV, GHSA), vendor-specific security advisories, and general web search results via the Tavily tool.
Ingestion points: Data enters the agent's context from vendor/distro advisories, CVE/CNA records, and ecosystem-specific sources. The instructions also explicitly suggest using web searches for discovery.
Boundary markers: The skill lacks explicit boundary markers, delimiters, or instructions for the agent to ignore potentially malicious instructions embedded within the external security records or search snippets.
Capability inventory: While the skill itself focuses on lookup and reporting, it instructs the agent to load related skills (poc-weaponization and deep-research-offensive) which may grant the agent access to more dangerous tools or command execution capabilities.
Sanitization: There are no requirements or procedures defined for sanitizing or validating external content before it is interpolated into the agent's reasoning process or final inventory reports.

cve-search