auto-mcm
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple shell commands and scripts (e.g., pipeline_manager.py, setup_workspace.py) to manage workspace status and pipeline transitions. These commands are executed silently in the background to streamline the user experience.
- [INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection via the extraction of text from user-provided PDF problem files. Ingestion points: PDF files at user-specified paths. Boundary markers: Not used during the initial configuration inference phase. Capability inventory: Shell script execution, file management, and Git access. Sanitization: The skill implements search query abstraction for external tools but lacks specific sanitization for ingested PDF text.
- [EXTERNAL_DOWNLOADS]: The skill automatically installs necessary Python packages (e.g., scipy, numpy, pandas, pdfplumber) from standard registries if they are missing from the environment.
- [DYNAMIC_EXECUTION]: The skill uses inline Python commands (python -c) for runtime dependency checks and data extraction from documents.
Audit Metadata