deployment-diagram
Pass
Audited by Gen Agent Trust Hub on Jul 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local shell scripts, specifically
scripts/new-canvasandscripts/tq, to create workspace files and interact with the tldraw application's API via curl. - [DATA_EXFILTRATION]: The skill accesses local configuration files (
server.json) located in application support directories (e.g.,~/Library/Application Support/tldraw/) to retrieve authentication tokens required for communicating with the local tldraw server. These tokens are only used for local API requests tolocalhost. - [EXTERNAL_DOWNLOADS]: The documentation provides a link to the official tldraw website (
https://offline.tldraw.com) to assist users in installing the necessary prerequisite software. - [PROMPT_INJECTION]: The skill parses user-provided prose or infrastructure lists to generate diagram components. This ingestion of untrusted data into a processing pipeline that generates executable canvas commands constitutes an indirect prompt injection surface, though the impact is restricted to the local drawing environment.
Audit Metadata