The Agent Skills Directory

[SAFE]: The skill connects to the official Open Targets Platform API at api.platform.opentargets.org. This is a well-known and trusted scientific resource used for therapeutic target identification.
[COMMAND_EXECUTION]: Uses a local Python helper script scripts/query_opentargets.py to manage API requests. This script is well-structured and provides a safe interface for executing both pre-defined and custom GraphQL queries.
[DATA_EXFILTRATION]: No evidence of unauthorized data transfer or sensitive file access. Network activity is strictly limited to the official Open Targets endpoint, and all query results are written to user-specified local paths via the --output flag.
[REMOTE_CODE_EXECUTION]: The skill's dependencies are limited to internal or standard libraries (e.g., science-skills-common). No patterns of downloading and executing untrusted remote code or dynamic code evaluation (like eval() or exec()) were found.
[SAFE]: While the skill ingests external scientific data from the API, which technically constitutes an indirect prompt injection surface, the data source is reputable and the skill handles the information as structured JSON, mitigating risks associated with untrusted content.

opentargets-database