find-untested-sources
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local analysis scripts written in C# and Python to scan the repository. These scripts perform read-only operations on source code to build an index of declarations and cross-reference them with test files.\n- [EXTERNAL_DOWNLOADS]: The analyzer relies on reputable packages like Microsoft.CodeAnalysis.CSharp and tree-sitter-language-pack. The skill instructions properly direct the agent to treat these as environment prerequisites rather than performing ad-hoc installations.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code from the repository. It mitigates this surface by performing structural parsing to extract only structural metadata like identifiers and namespaces.\n
- Ingestion points: Source files are read in
scripts/Find-UntestedSources.csandscripts/find_untested_sources.py.\n - Boundary markers: The output is limited to a structured JSON object.\n
- Capability inventory: The skill scripts only have file-read and stdout-write capabilities; they do not perform network operations or execute untrusted data.\n
- Sanitization: The use of formal parsers ensures that only specific symbol tokens are extracted from the input files.
Audit Metadata