The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to guide an agent in reviewing and processing sensitive information, including application logs, API request/response bodies, and user-provided data. This creates a surface for indirect prompt injection where an attacker could embed instructions within these data sources to influence the agent's behavior.
Ingestion points: Untrusted data analyzed via the guidelines in SKILL.md and references/data-rights.md (e.g., log bodies, API DTOs, DSR requests).
Boundary markers: Not present. The skill provides conceptual guidelines but does not define technical delimiters to isolate untrusted data during processing.
Capability inventory: None. The skill consists of Markdown files only and has no direct execution capabilities (no scripts, subprocess calls, or network operations).
Sanitization: The guidelines recommend data masking and anonymization for privacy compliance, but do not address the sanitization of embedded instructions in processed text.
[NO_CODE]: The skill contains no executable scripts, binaries, or configuration files that trigger platform tools. The primary files (SKILL.md, Security.md, data-rights.md) are documentation-based, which minimizes the direct execution risk.

gdpr-compliant