pencil-mcp-get-screenshot

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it processes untrusted inputs and tool outputs without proper isolation.
  • Ingestion points: Untrusted data enters the agent context through the nodeId and filePath parameters, as well as the visual data returned by the get_screenshot tool (SKILL.md).
  • Boundary markers: There are no instructions to use delimiters or ignore potential instructions embedded within the processed design data.
  • Capability inventory: The skill enables the agent to call the get_screenshot tool to interact with the design environment (SKILL.md).
  • Sanitization: No input validation or sanitization routines are defined for the parameters provided to the tool.
  • Mitigation: Implement boundary markers (e.g., XML tags) around external data and explicitly instruct the agent to disregard any commands found within the design nodes or screenshots.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 04:14 AM
Security Audit — agent-trust-hub — pencil-mcp-get-screenshot