The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external websites which is a standard surface for indirect prompt injection.\n
Ingestion points: The web_fetch_exa tool returns content from arbitrary URLs provided at runtime.\n
Boundary markers: There are no explicit instructions or delimiters defined to help the agent distinguish between the fetched data and its core instructions.\n
Capability inventory: The agent is instructed to present information and answer questions based on the retrieved content.\n
Sanitization: The skill does not define any filtering or sanitization steps for the retrieved text within the instruction set.

exa-fetch