pptx-deck-creation
Warn
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill provides functional Python code snippets within its reference documentation (e.g., in
visual-asset-adapters.md) and directs the agent to run them. These snippets handle tasks such as downloading files, searching icon databases, and interacting with external image generation APIs. - [EXTERNAL_DOWNLOADS]: The skill requires the agent to perform network requests to external domains not included in the trusted vendor list, specifically
getdesign.mdfor design profiles andapi.iconify.designfor icon assets. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data from user-provided reference PPTX files and design specifications fetched from the web, which constitutes an attack surface for indirect prompt injection.
- Ingestion points: Reference PPTX files (analysed via OOXML parsing) and catalog entries from
getdesign.md. - Boundary markers: The skill contains explicit instructions to ignore commands or instructions embedded within the design references.
- Capability inventory: The skill enables file read/write operations, network requests via Python libraries, and the generation/execution of scripts.
- Sanitization: It provides guidelines for secure XML parsing to mitigate risks during the inspection of untrusted PPTX packages.
Audit Metadata