import-screenshot-to-prototype
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its core function of processing external data.
- Ingestion points: User-provided screenshots or image references (defined in SKILL.md).
- Boundary markers: Absent. There are no instructions to the agent to ignore or sanitize embedded instructions within the images.
- Capability inventory: The skill has 'fs:write' and 'file-write' capabilities (defined in open-design.json), allowing it to create files based on processed data.
- Sanitization: Absent. The workflow does not specify any validation or filtering of content extracted from the images.
- [PROMPT_INJECTION]: The metadata in
open-design.jsonexplicitly requests a capability namedprompt:inject. While this may be a platform-specific configuration for providing skill instructions, the terminology is highly non-standard and suggests a design intended to manipulate the agent's instruction stream.
Audit Metadata