break
Warn
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill requires the agent to build a temporary 'harness page' and a new 'scratch route' within the existing application codebase. This process involves generating code that imports and executes local project components with synthetic data at runtime.
- [INDIRECT_PROMPT_INJECTION]: The skill functions by processing untrusted data sources (identified in 'scenarios.md' as user input, CMS content, and API data) to generate test scenarios. This represents a vulnerability surface where the agent ingests and renders content without specific sanitization or boundary markers.
- Ingestion points: scenarios.md (Content length, Content shape, and Quantity axes).
- Boundary markers: None identified; the skill does not instruct the agent to use delimiters or sanitization when rendering these scenarios.
- Capability inventory: The agent performs file-system writes to create the harness and potentially uses local environment capabilities to render the page.
- Sanitization: None identified; the instructions focus on rendering raw data 'untouched' to observe breaks.
- [COMMAND_EXECUTION]: The agent is directed to perform file system modifications, including creating temporary routes, fixture files, and harness components within the user's project directory to facilitate the visual testing environment.
Audit Metadata