The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes academic data from OpenAlex, which creates a surface for indirect prompt injection. This is addressed by specific guardrail instructions for the agent.
Ingestion points: File SKILL.md identifies that functions like openalex_find_works return payload text.
Boundary markers: The skill documentation explicitly warns the agent to treat retrieved text as untrusted and ignore embedded instructions.
Capability inventory: SQL queries and embedding requests are performed via curl in SKILL.md.
Sanitization: Input validation is mandated for all entity IDs using regular expressions.
[SAFE]: All external communications target the vendor's own API infrastructure (api.exopriors.com), which is standard behavior for the skill's purpose.

openalex