The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses curl to interact with a local HTTP API at localhost:7236 to read and modify canvas data.
[REMOTE_CODE_EXECUTION]: The skill documentation describes an exec endpoint (/api/doc/:id/exec) that enables the execution of arbitrary JavaScript code within the tldraw environment.
[PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection (Category 8) because it processes data from the tldraw canvas which may be controlled by an external party or the user.
Ingestion points: The agent reads canvas state and shape information using the GET /api/doc/:id/shapes endpoint.
Boundary markers: No boundary markers or 'ignore' instructions are provided to distinguish between canvas data and agent instructions.
Capability inventory: The agent can perform file writes (e.g., saving screenshots to /tmp/canvas.jpg) and execute arbitrary JavaScript via the exec API.
Sanitization: There is no evidence of sanitization or validation of the text content found within shapes before it is processed by the agent.

tldraw-canvas