stata-recipes

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill serves as a documentation and code-generation guide for Stata tasks. It focuses on helping users avoid common statistical errors and provides structured workflows for complex modeling.
  • [DATA_EXFILTRATION]: No evidence of sensitive data access or exfiltration. The skill actively encourages safe data handling practices, such as treating raw data files as read-only and using new filenames for cleaned datasets to prevent data loss.
  • [COMMAND_EXECUTION]: The skill generates Stata and Python code for user execution. It utilizes standard libraries like sfi for in-process interop and provides instructions for batch processing. Examples include standard commands like subprocess.run and Stata's shell command, used appropriately for statistical workflows.
  • [EXTERNAL_DOWNLOADS]: Mentions the installation of well-known community-contributed Stata packages (such as estout and kpss) from the standard Statistical Software Components (SSC) archive via the ssc install command.
  • [INDIRECT_PROMPT_INJECTION]: While the skill ingests user descriptions to produce code, it incorporates diagnostic steps and validation checks (like unit-root testing) that act as safeguards against generating invalid or misleading analytical scripts. It provides clear structural templates for the resulting code and follows established econometric best practices.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 05:49 PM
Security Audit — agent-trust-hub — stata-recipes