windows-agent-shell
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/check_text.pyusessubprocess.runto executegit status. This is done to identify modified or untracked files in a repository when the--changedflag is used. The implementation uses a list of arguments withoutshell=True, which prevents shell injection vulnerabilities. - [INDIRECT_PROMPT_INJECTION]: The skill provides the
scripts/check_text.pytool which reads local file content to detect mojibake, BOMs, and line-ending inconsistencies. While this ingests untrusted data from the local file system, the script performs only passive byte-level and encoding analysis and does not interpret or execute the file content as instructions. - [SAFE]: The skill's primary purpose is educational and diagnostic. It provides specific receipts and scripts (
scripts/probe_argv.pyandscripts/check_text.py) to troubleshoot shell argument splitting and text corruption issues. No network operations, credential harvesting, or persistence mechanisms were found.
Audit Metadata