The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the ClinPGx REST API (api.clinpgx.org) and processes it into the agent's context. This creates a surface for indirect prompt injection if the external data source were to contain malicious instructions.
Ingestion points: API responses from multiple endpoints (gene, chemical, guideline, etc.) are parsed as JSON and returned to the agent.
Boundary markers: The provided scripts do not show explicit delimiters or warnings to the LLM to ignore instructions embedded within the biological data.
Capability inventory: The skill has the ability to perform network GET requests and write to the local filesystem for caching purposes.
Sanitization: Data is parsed using standard JSON libraries, which prevents typical code injection but does not filter natural language instructions.
[EXTERNAL_DOWNLOADS]: The skill communicates with api.clinpgx.org to retrieve pharmacogenomic information, guidelines, and allele data. This is the primary function of the skill and uses standard HTTP GET requests.
[DATA_EXPOSURE_AND_EXFILTRATION]: The script scripts/query_clinpgx.py implements a caching mechanism (cached_query) that writes API responses to local files (e.g., cyp2d6_cache.json). This is a standard performance optimization and does not appear to target sensitive user files or credentials.

clinpgx-database