The Agent Skills Directory

[DATA_EXFILTRATION] (LOW): The documentation provides examples of network requests to api.platform.opentargets.org using the requests library. While this is a legitimate scientific API, it is not on the trusted domain whitelist, which qualifies as a low-severity data exposure finding in the absence of sensitive data access.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill implements an ingestion point for external data from a GraphQL API. 1. Ingestion points: GraphQL API responses from Open Targets. 2. Boundary markers: None present in code examples. 3. Capability inventory: requests.post for data retrieval; no write or execute capabilities detected. 4. Sanitization: None described. As the data is primarily used for biological entity retrieval and reasoning, the risk remains low.
[SAFE] (SAFE): No malicious instructions, obfuscation, privilege escalation, or persistence mechanisms were detected in the files.

opentargets-database