The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from a marketplace of over 60 external providers, creating a surface for indirect prompt injection if the data contains malicious instructions.
Ingestion points: Data returned by qveris_execute and external files retrieved via full_content_file_url (SKILL.md, references/rest-api.md).
Boundary markers: No instructions for boundary markers or instruction-ignoring delimiters are provided for external data.
Capability inventory: The skill uses network-enabled tools for data retrieval and allows fetching external content from signed URLs.
Sanitization: There is no mention of sanitizing or validating the integrity of the payloads before they are processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill defines a process for fetching large data payloads from dynamically generated signed URLs.
Evidence: The skill instructions state "If the full JSON is required, fetch that signed URL and use the full payload" in SKILL.md.

qveris