The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external URLs via the website_url input and user-supplied strings via the policy_text input. This content is then processed using NLP capabilities, creating a surface for indirect prompt injection attacks where malicious instructions hidden in the website or policy could override the agent's intended behavior.\n- Ingestion points: External content is fetched from website_url and user input is taken from policy_text as defined in skill.yaml.\n- Boundary markers: The documentation does not mention the use of delimiters or specific instructions for the AI to ignore embedded commands within the analyzed text.\n- Capability inventory: The skill is capable of web crawling (scripts/cookies), local file access (via source_code_path), and NLP-based extraction.\n- Sanitization: There is no evidence of sanitization or filtering for the external data before it is passed to the analysis engine.

gdpr-ccpa-privacy-auditor