working-on
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill designates a tracking document as the persistent source of truth for an agent's task, creating a surface where content stored in the file could influence the agent's behavior or decisions in future sessions.\n
- Ingestion points: The tracking document located at the file path provided in the
$pathargument.\n - Boundary markers: The instructions do not provide specific delimiters or ignore-instructions to isolate the tracking document's content from the agent's core operational logic.\n
- Capability inventory: The skill assumes capabilities for file system interaction (read, write, create) and git repository management.\n
- Sanitization: No content sanitization, validation, or filtering is defined for the document's contents.
Audit Metadata