tripo-studio
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill generates and injects a custom JavaScript helper script (
studio_helpers.js) into the browser context. This script is used to facilitate reliable interaction with complex web components (Radix UI) that do not respond to standard click events. - [INDIRECT_PROMPT_INJECTION]: The skill reads and processes dynamic content from the Tripo Studio web interface, such as tooltips, labels, and model metadata. This represents a potential surface where adversarial text on the website could influence agent behavior.
- Ingestion points: Web page content from
studio.tripo3d.ai(tooltips, UI labels, asset names). - Boundary markers: The instructions provide specific mapping commands (
T.map()) but do not define explicit delimiters to separate web content from agent instructions. - Capability inventory: The skill utilizes
javascript_tool,file_upload, and automated browser navigation. - Sanitization: There is no evidence of sanitization or filtering of text extracted from the web interface before processing.
- [COMMAND_EXECUTION]: The skill uses browser tools to navigate to external URLs and perform file uploads of local reference images to the Tripo Studio service.
Audit Metadata