ux-flows
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute local Python scripts (
docs/ux/lint.py,plugins/super-ux/scripts/bp_index.py) to perform project validation and index regeneration. These scripts are described as being 'seeded' or managed by the skill pack. - [EXTERNAL_DOWNLOADS]: The skill recommends installing the
ssheleg/task-pipelineplugin from an external marketplace to automate the execution of UX plans. This involves downloading and executing third-party code, although the plugin is associated with the skill's author (ssheleg). - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by ingesting and analyzing data from external reference sources (e.g., Refero, Mobbin, Lazyweb) and competitor funnel research. It includes a specific mitigation instruction advising the agent to treat fetched references as data rather than instructions.
- [DYNAMIC_EXECUTION]: The skill framework involves the use of dynamically managed or generated scripts for maintenance and quality checks, which are executed during the workflow.
Audit Metadata