general-plot-digitizer
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/digitize_pipeline.pyusessubprocess.run()to execute internal scripts likeextract_metadata.py. These calls use structured argument lists and the current interpreter, which is a secure method for process orchestration. - [EXTERNAL_DOWNLOADS]: The skill connects to OpenAI and Google Gemini APIs for image analysis. These connections target established service providers and are essential for the tool's scientific purpose.
- [PROMPT_INJECTION]: The skill ingests untrusted data from images and VLM responses. 1. Ingestion points: VLM responses are processed in
scripts/extract_metadata.pyand initial inspection phases. 2. Boundary markers: No explicit delimiters are used to separate external data in prompt templates. 3. Capability inventory: The skill can execute local Python scripts viasubprocess.runand write data files. 4. Sanitization: Filenames are sanitized via regex indigitize_pipeline.pyto prevent path traversal. The risk is low as the pipeline is designed for deterministic data extraction.
Audit Metadata