code-data-analysis-scaffolds

Pass

Audited by Gen Agent Trust Hub on Apr 16, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [SAFE]: No malicious patterns or security vulnerabilities were identified. The skill acts as a set of instructional templates for professional technical workflows.\n- [PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted external data for analysis purposes, which is its primary function.\n
  • Ingestion points: Guidelines in template.md and eda-customer-churn.md involve reading user-provided files (e.g., CSV) into the agent context.\n
  • Boundary markers: The templates do not currently instruct the agent to use delimiters or protective instructions when processing external data.\n
  • Capability inventory: The skill encourages operations involving file system interaction and shell execution for testing (e.g., pytest).\n
  • Sanitization: The scaffolds focus on data quality and statistical validity rather than sanitizing content against instruction-like strings.\n- [COMMAND_EXECUTION]: The documentation and examples suggest using standard command-line tools such as pytest, mutmut, and great_expectations for code and data validation.\n- [EXTERNAL_DOWNLOADS]: The skill mentions the installation of well-known open-source libraries like umap-learn and mutmut in its advanced methodology documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 16, 2026, 02:28 AM