class-diagram
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (prose, class lists, or source code) and uses it to generate JavaScript commands that are executed within the tldraw application's environment. This represents a potential injection surface if malicious names or attributes are provided in the source description.\n
- Ingestion points: Input described in the Workflow section of SKILL.md.\n
- Boundary markers: Absent in the parsing and rendering instructions.\n
- Capability inventory: Arbitrary JS execution via scripts/tq calling the /exec endpoint of the local tldraw server.\n
- Sanitization: No sanitization or validation logic for class names or member descriptions is specified in the workflow.\n- [DYNAMIC_EXECUTION]: The skill's core functionality relies on generating and executing JavaScript at runtime to manipulate the tldraw canvas, as documented in references/shared/bridge.md.\n- [COMMAND_EXECUTION]: The skill includes shell scripts (scripts/tq and scripts/new-canvas) that execute local commands like curl, python3, and open to manage files and communicate with the local tldraw server.
Audit Metadata