pmg-review-validation

Pass

Audited by Gen Agent Trust Hub on Nov 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a command-line utility named pmg for validating and repairing presentation artifacts. This involves executing external binaries with parameters provided in the instructions.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external PowerPoint files (input.pptx) to identify and resolve audit defects, which creates an attack surface for indirect prompt injection.
  • Ingestion points: The skill ingests data from external presentation files passed to the pmg command.
  • Boundary markers: The skill defines safety classifications (e.g., SAFE_AUTOMATIC, REQUIRES_REVIEW) to limit the scope of automated changes, though it does not explicitly mention markers to prevent the agent from following instructions embedded within the slide content.
  • Capability inventory: The skill possesses capabilities to execute shell commands and write modified files to the local file system.
  • Sanitization: While the skill implements "safety classification" logic to gate specific types of repairs, it lacks explicit content sanitization or filtering for potential prompt injection payloads hidden within the presentation data (text, notes, or metadata).
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 12, 2026, 04:39 PM