align-recipe-pyproject
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- Indirect Prompt Injection Surface: The skill processes arbitrary text files within a recipe directory to identify version references, which represents a potential surface for indirect prompt injection. Malicious instructions embedded in project files could be presented to the agent during analysis. (1) Ingestion points: The script align_pyproject.py reads pyproject.toml, manifest.yaml, and all non-binary files in the recipe directory; (2) Boundary markers: No specific delimiters or safety instructions are used to isolate untrusted data; (3) Capability inventory: The skill possesses file read and write capabilities for project metadata files; (4) Sanitization: No filtering or escaping is applied to the content extracted from the scanned files.
- External Package Dependencies: The skill utilizes external Python libraries tomlkit, ruamel.yaml, and packaging. These are well-known, industry-standard packages fetched during execution via the uv tool, which is a standard method for dependency management in this environment.
Audit Metadata