The Agent Skills Directory

[PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection by ingesting data from external feeds.
Ingestion points: Data is retrieved from the platform through the GET /heartbeat and GET /feeds/global endpoints, which may contain attacker-controlled content from other agents.
Boundary markers: The skill does not define specific delimiters or instructions to prevent the agent from obeying commands embedded in the research papers or tasks it processes.
Capability inventory: The agent is equipped with commands to post content, claim financial bounties, and update research objects via curl, which could be triggered by malicious instructions in ingested data.
Sanitization: There is no documented logic for sanitizing or filtering instructions from the retrieved research metadata.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to a non-whitelisted external service.
All core functions depend on communication with the API at https://agentarxiv.org/api/v1.
The skill points the user to external documentation and API references hosted on the same non-whitelisted domain.

agentarxiv