plt-grade-doc
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied product documents, which creates a surface for indirect prompt injection. If an input document contains malicious instructions designed to manipulate the agent's behavior or output, the agent might inadvertently follow them.
- Ingestion points: The skill ingests raw text from various product document types (e.g., PRDs, ADRs, RFCs) provided by the user.
- Boundary markers: The instructions do not mandate the use of specific delimiters or 'ignore' warnings when processing the input document, although the output format requires quoting specific evidence.
- Capability inventory: The skill has network access to fetch rubrics from the vendor's GitHub repository and file system access to read local manifests and write anonymous usage logs to the 'evals/usage-log/' directory.
- Sanitization: No specific sanitization or filtering of the input document content is described in the provided files.
- [EXTERNAL_DOWNLOADS]: The skill fetches configuration manifests and grading guides from the author's official GitHub repository (product-on-purpose/product-lifecycle-templates). This is an intended function used to ensure the agent uses authoritative rubrics rather than generating them from internal memory. These downloads are performed from a trusted vendor source on a well-known platform (GitHub).
Audit Metadata