The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from Large Language Models (LLMs), which constitutes a potential vulnerability surface for indirect prompt injection if the model output is influenced by malicious data.
Ingestion points: Data enters the system via the genai_response object and response_text variable in the code examples provided in SKILL.md.
Boundary markers: The provided implementation examples do not use explicit delimiters or specific system instructions to isolate model output from the agent's internal control logic.
Capability inventory: The skill includes patterns for executing database operations (db.execute) based on the ingested model content.
Sanitization: The skill mitigates these risks by providing a validate_medical_response function that checks for hallucination markers, structural integrity, and potential PII leaks using regular expression patterns.

genai-services