saylog-manual-authoring
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to clone documentation and context from the vendor's GitHub repositories (e.g., github.com/douinc/saylog/wiki) to ensure manuals use correct terminology and policies. These are official vendor resources used for reference.
- [COMMAND_EXECUTION]: The skill automates manual creation through several scripts. A shell script (render.sh) invokes Google Chrome in headless mode to convert HTML to PDF, while Python scripts handle image annotation (annotate.py), image normalization (normalize_shot.py), and PDF-to-PNG conversion for verification (verify.py).
- [PROMPT_INJECTION]: The skill processes external data from user-supplied screenshots and text, as well as documentation from git repositories. While this creates a surface for indirect prompt injection where instructions could be embedded in the data, the risk is low given the specific manual-authoring use case and the use of official vendor documentation.
Audit Metadata