The Agent Skills Directory

[SAFE]: No malicious patterns or critical vulnerabilities were detected. The skill uses standard network libraries to communicate with a search service, with no evidence of unauthorized data exfiltration or credential harvesting.\n- [PROMPT_INJECTION]: The skill presents an inherent surface for indirect prompt injection due to its core function of retrieving external web content.\n
Ingestion points: The scripts/search.py script retrieves search results from external sources via the configured SearXNG instance.\n
Boundary markers: There are no explicit boundary markers or instructions in the skill's prompts to help the agent distinguish between its system instructions and the content of search results.\n
Capability inventory: The skill uses allowed-tools: shell in SKILL.md to execute the Python search utility, which is appropriate for its function.\n
Sanitization: While the script extracts specific fields (title, URL, snippet) to provide some structure, it does not perform content-level filtering or sanitization on the retrieved text, which is typical for search tools.

searxng-search