edrawmax-diagram
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from an external API endpoint and instructs the agent to interpolate response fields into a shell command line.
- Ingestion points: API responses from
https://api.edrawmax.cn/api/ai/skills/generate-{type}containingpng_urlandsvg_urlinSKILL.md. - Boundary markers: Absent within the command string interpolation template.
- Capability inventory: Shell execution of the download script (
download_diagram.py) via the command line interface. - Sanitization: The
download_diagram.pyscript strictly validates that input URLs use HTTPS and belong to trusted EdrawMax/Wondershare domains before performing the download, though the shell-level interpolation happens prior to script execution. - [COMMAND_EXECUTION]: In
SKILL.mdStep 3, the agent is instructed to run a local Python script (download_diagram.py) to download and store files on the local filesystem.
Audit Metadata