open-pptd

Pass

Audited by Gen Agent Trust Hub on Oct 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches pre-compiled binary assets and installation scripts from the vendor's GitHub repository (shiquda/open-pptd-skill) during the setup process.
  • [REMOTE_CODE_EXECUTION]: The installation process involves downloading and executing an installation script from GitHub. Additionally, the skill installs and runs a native binary compiler (pptd) to handle core presentation tasks.
  • [COMMAND_EXECUTION]: Orchestrates the pptd compiler and system tools such as LibreOffice and pdftoppm using subprocess.run to perform PDF conversion and image rasterization.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection when processing untrusted presentation data.
  • Ingestion points: Ingests data from external .pptx and .pptd files via scripts/convert_pptx.py, scripts/export_pptx.py, and scripts/validate_pptx.py.
  • Boundary markers: Relies on YAML structure for configuration but lacks explicit markers or instructions to prevent the agent from following commands embedded within slide content.
  • Capability inventory: Includes shell command execution, file system operations, and network access (for installation) across its script library.
  • Sanitization: Content extracted from source presentations is processed without comprehensive sanitization before being presented to the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 19, 2026, 06:29 PM