canva-bulk-create

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Potential for indirect prompt injection from untrusted data sources.
  • Ingestion points: The skill processes data from uploaded files, pasted text, and remote URLs as specified in Step 1 of SKILL.md.
  • Boundary markers: There are no explicit markers or instructions provided to the agent to disregard or ignore potentially malicious commands embedded within the tabular data rows.
  • Capability inventory: The skill uses Canva:autofill-design and Canva:upload-asset-from-url to generate content based on the input data (SKILL.md, Step 5).
  • Sanitization: No data validation, escaping, or sanitization steps are defined before interpolating external values into the design creation payload.
  • [EXTERNAL_DOWNLOADS]: The skill is designed to fetch tabular data and assets from user-provided URLs.
  • Evidence: Step 1 allows the agent to "fetch the resource and parse the response as tabular data" from a URL. Additionally, Step 4 and 5 describe a workflow using Canva:upload-asset-from-url to process image URLs found in the data source.
  • Context: While this functionality is central to the skill's purpose (bulk creating from external data), it allows the agent to perform network requests to arbitrary external endpoints provided by the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 05:41 PM
Security Audit — agent-trust-hub — canva-bulk-create