json-canvas
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a structure for processing
.canvasfiles which can contain text nodes with Markdown. This constitutes a potential attack surface where an agent might ingest data containing malicious instructions if the file source is untrusted. - Ingestion points: JSON fields including
text,label,file, andurlwithin the.canvasfile structure. - Boundary markers: The JSON schema format provides implicit field boundaries.
- Capability inventory: The skill enables the agent to read, parse, and generate files based on the specified format.
- Sanitization: No specific instructions are provided for sanitizing or escaping the content of text nodes during processing.
- [EXTERNAL_DOWNLOADS]: The skill contains references to the official JSON Canvas project website and its repository on GitHub. These are well-known technology resources used for format verification and documentation.
Audit Metadata