json-canvas
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill enables the agent to interact with .canvas files that may contain untrusted Markdown text, creating a potential vector for indirect prompt injection.\n
- Ingestion points: Data is ingested from the 'text' field of nodes within .canvas files as described in SKILL.md.\n
- Boundary markers: Absent. The instructions do not define specific delimiters to isolate the content of text nodes from the agent's operating context.\n
- Capability inventory: The skill's primary capabilities include reading, creating, and modifying files on the local filesystem.\n
- Sanitization: Absent. The specification does not include measures for sanitizing or validating text content to prevent the execution of embedded instructions.\n- [EXTERNAL_DOWNLOADS]: The skill provides informational links to the JSON Canvas specification and its official GitHub repository.\n
- The skill references documentation hosted at jsoncanvas.org and source code at the obsidianmd/jsoncanvas GitHub repository which is a well-known project.
Audit Metadata