sketchbook
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from the project codebase and a local configuration file to influence its code generation logic, creating a surface for indirect instructions.
- Ingestion points: The agent is instructed to read
sketchbook/taste.md(Rule 2) and the broader project stack including framework, styling, and animation library files (Rule 1). - Boundary markers: The instructions do not specify the use of delimiters or specific warnings to ignore instructions embedded within the files being read.
- Capability inventory: The skill has the capability to write files (Rule 6:
sketchbook/taste.mdandsketchbook/kept/), modify project routes, and start/open a local dev server (Rule 5). - Sanitization: There is no evidence of sanitization or filtering of the content read from
taste.mdbefore it is used to generate new component code. - [DYNAMIC_EXECUTION]: The skill generates UI components (hover, toggle, drag, etc.) at runtime and executes them within the project's development environment.
- The agent generates interaction code using the project's local stack (Rule 3).
- The agent is instructed to start a dev server and open a browser path (
/sketchbook) to execute and verify the generated code (Rule 5).
Audit Metadata