building-remix-design-system
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the user to implement a token pipeline that imports external npm packages as executable modules within the tool/extract_tokens.mjs script. It also leverages dart run build_runner for automated code generation during the component creation phase.\n- [COMMAND_EXECUTION]: The development workflow requires executing various shell commands for package scaffolding (Melos), building components, and running verification scripts in CI environments.\n- [EXTERNAL_DOWNLOADS]: The skill is configured to retrieve design assets and token data from external platforms such as the Figma API and public npm registries.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted data from external design sources. Ingestion points: Data enters the context via Figma API responses, JSON token files, PDF brand books, and screenshots (references/source-extraction.md). Boundary markers: The pipeline implements deterministic stages and requires committed snapshots to detect drift, but lacks explicit instructions for the agent to ignore linguistic commands embedded in design data. Capability inventory: The skill facilitates writing to the local filesystem and executing scripts in the tool/ directory. Sanitization: Implementation includes validation of parsed values, such as hex color string verification and numeric NaN checks (references/token-pipeline.md).
Audit Metadata