codex-ppt
Pass
Audited by Gen Agent Trust Hub on May 29, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs well-known Python dependencies (python-pptx, Pillow, openai) from official registries during its bootstrap process.
- [COMMAND_EXECUTION]: Uses subprocess calls to manage its local virtual environment and verify package availability. All executed commands use paths derived from the local environment or static code strings.
- [CREDENTIALS_UNSAFE]: API keys are stored in a local .env file within the user's home directory. The skill explicitly sets file permissions to 0600 (user read/write only) to protect these secrets.
- [DATA_EXFILTRATION]: Network activity is restricted to calling the configured OpenAI-compatible image generation API. A connectivity check is performed against the official or user-defined models endpoint.
- [PROMPT_INJECTION]: The skill ingests user-provided text (articles, reports) to generate slide content. While this presents a surface for indirect prompt injection, the risk is inherent to the skill's primary function, and the workflow uses structured JSON job files to isolate external content within the generation pipeline.
Audit Metadata