The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, third-party platforms (Reddit, Twitter, YouTube, Web search), which creates a potential surface for indirect prompt injection attacks.
Ingestion points: Data is retrieved from external sources using tools like reddit_app_dynamic_search, youtube_web_v2_general_search, twitter_web_search_timeline, exa_search, and tavily_search listed in SKILL.md.
Boundary markers: There are no explicit instructions for the agent to treat retrieved content as untrusted or to use specific delimiters to isolate potentially malicious embedded instructions.
Capability inventory: The skill possesses the capability to perform wide-ranging searches across multiple platforms using sandbase_call_tool.
Sanitization: The instructions do not include steps for sanitizing, filtering, or validating the content returned from external tool calls.

content-ideation