elite-powerpoint-designer

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions direct the agent to execute local Python scripts (scripts/analyze_content.py and scripts/validate_consistency.py) to perform its primary functions. These scripts are invoked via the shell to process input data and validate output files.
  • [EXTERNAL_DOWNLOADS]: The skill requires several external dependencies from standard registries, including Python packages (python-pptx, pillow, pyyaml) and a specific Node.js package (@gongrzhe/office-powerpoint-mcp-server) via the Smithery CLI.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied markdown content to determine the presentation's structure, layout, and visual mapping, which presents a surface for indirect prompt injection.
  • Ingestion points: Markdown files (input.md) and user-provided text content.
  • Boundary markers: The instructions lack explicit boundary markers or delimiters to separate user content from design instructions.
  • Capability inventory: The skill has the capability to execute shell scripts and perform file system operations via the Office-PowerPoint-MCP-Server.
  • Sanitization: There is no mention of sanitization, filtering, or validation of the input markdown to prevent embedded instructions from influencing agent behavior beyond the intended design mapping.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 08:37 AM
Security Audit — agent-trust-hub — elite-powerpoint-designer