product-boot

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes project decision records from local files, which creates an ingestion surface for data that could potentially contain embedded instructions.
  • Ingestion points: The skill reads from docs/adlc/memory/pdr/pdr.md, .adlc/memory/pdr/pdr.md, and individual PDR files in docs/adlc/memory/pdr/ and .adlc/drafts/pdr/.
  • Boundary markers: Record data is structured into markdown tables under the ## PDR Context and ## Drafts Pending Review headers, providing structural separation from other session instructions.
  • Capability inventory: The skill has local file read capabilities and write access to the .adlc/drafts/pdr/ directory for record capture.
  • Sanitization: No specific sanitization or filtering of record content is described, though the skill strictly adheres to a table-based output contract.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 01:39 PM