json-canvas
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes existing .canvas files which provides a surface for indirect prompt injection via text or label fields.
- Ingestion points: SKILL.md describes workflows for reading and parsing existing .canvas files.
- Boundary markers: The instructions lack delimiters or warnings to ignore instructions embedded in node content.
- Capability inventory: SKILL.md outlines capabilities for writing updated JSON back to the file system.
- Sanitization: There is no evidence of sanitization or validation of the content within node fields to prevent injection.
Audit Metadata