ui-convert-token-miner
Pass
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The main execution script
scripts/mine.tsspawns Python subprocesses usingchild_process.execFileto run its internal helper scriptsxaml-extractor.pyanddart-extractor.py. This is a standard architectural choice for handling multi-language parsing requirements. - [PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from the project being analyzed (source code style definitions).
- Ingestion points: Reads arbitrary CSS, SCSS, LESS, JS, TS, XAML, and Dart files defined in the project's index.
- Boundary markers: None; the skill relies on regex and XML parsing to extract values.
- Capability inventory: The skill has the ability to spawn subprocesses and write results to
tokens.jsonon the local file system. - Sanitization: Values are normalized into canonical formats (e.g., hex colors, pixel numbers), providing a layer of data validation before the tokens are saved.
Audit Metadata