canva-presentation

Pass

Audited by Gen Agent Trust Hub on Apr 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute specific Python scripts (e.g., auth_check.py, get_presentation.py, update_slide.py) to interact with the Canva API. These operations are restricted to the scripts provided within the skill environment.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing arbitrary text content retrieved from external Canva presentations.
  • Ingestion points: Data enters via extract_all_text.py, get_presentation.py, and get_slide.py (SKILL.md) when reading Canva design data.
  • Boundary markers: The instructions mandate a 3-mode workflow (PLAN, CLARIFY, IMPLEMENT) which requires the agent to plan and confirm all changes with the user, serving as a procedural boundary.
  • Capability inventory: Access to the Bash tool for script execution and file system tools (Read, Write, Edit, AskUserQuestion).
  • Sanitization: There is no evidence of automated content sanitization or escaping for the presentation data before it is presented to the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 20, 2026, 12:43 AM
Security Audit — agent-trust-hub — canva-presentation