drawio-aws

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation advises the user to install the drawio-ai-kit from the vendor's repository (github:sparklabx/drawio-ai-kit) via npm. This is a standard dependency for the tool's operation.
  • [DYNAMIC_EXECUTION]: The workflow involves generating or editing a JavaScript file (build.mjs) and executing it with node to perform diagram layout and validation. This is a core feature of the declarative diagramming engine.
  • [INDIRECT_PROMPT_INJECTION]: Verbatim user requests and clarifications are passed into a subagent prompt, creating a potential surface for indirect injection.
  • Ingestion points: The user's diagram request and clarifications are interpolated into the subagent prompt in SKILL.md.
  • Boundary markers: The instructions use structured text blocks and distinct numbered steps to guide the subagent's behavior.
  • Capability inventory: The subagent is authorized to run specific CLI commands (drawio-ai), execute node scripts (build.mjs), and perform vision-based self-checks on rendered images.
  • Sanitization: There is no explicit sanitization or escaping of the user-provided content before it is passed to the subagent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 01:53 PM
Security Audit — agent-trust-hub — drawio-aws