The Agent Skills Directory

[SAFE]: The skill demonstrates safe behavior by restricting its tool usage to the designated zai-web-search utility and providing instructions focused on objective synthesis and source verification. No credentials, obfuscation, or dangerous execution patterns were found.- [PROMPT_INJECTION]: While the skill ingests untrusted data from web search results, which is a surface for indirect prompt injection, it lacks high-privilege capabilities such as file system writes or network exfiltration tools. The ingestion points are search snippets processed in SKILL.md; boundary markers are absent; the capability inventory is limited to search; and no explicit sanitization is defined. This is considered a low-risk exposure inherent to the skill's primary research purpose.

search-expert