sn-ppt-creative
Pass
Audited by Gen Agent Trust Hub on May 27, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes Python one-liners and standalone scripts to process metadata and organize slides. These commands are used to invoke the model client and bundle generated images into a PPTX file. All shell commands are static and operate within the workspace.
- [EXTERNAL_DOWNLOADS]: The skill declares a dependency on 'python-pptx', an established library for presentation management. This is a standard and safe dependency for the intended task.
- [DATA_EXFILTRATION]: Internal scripts were reviewed for network activity. No unauthorized exfiltration or unexpected network connections were found; data flow is confined to the local deck directory.
- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it incorporates user-supplied queries and document digests into generative prompts. Ingestion points: Untrusted data is retrieved from task_pack.json and info_pack.json. Boundary markers: System prompts use structured Markdown headers to separate instructions from data. Capability inventory: The skill can execute local Python logic and call external image generation tools. Sanitization: It includes 'sanitize_prompt.py', which filters technical metadata (like hex codes and CSS) to prevent image rendering artifacts, providing a layer of content control.
Audit Metadata