The Agent Skills Directory

[SAFE]: The skill's functionality and example commands are consistent with its stated purpose and do not contain any detected malicious patterns or suspicious data handling.
[PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it ingests untrusted data from web searches while maintaining shell execution capabilities. * Ingestion points: Content fetched via google_web_search. * Boundary markers: Absent. * Capability inventory: Access to run_shell_command, read_file, and google_web_search tools. * Sanitization: No sanitization or validation of ingested web content is explicitly described.

deep-research-swarm