jan-project-standard

Pass

Audited by Gen Agent Trust Hub on Oct 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes Python scripts that use subprocess.run to manage project initialization and Git configuration. Evidence: new_project.py and setup_git_identity.py execute commands such as git init, uv sync, and ssh-keygen. Analysis: All command executions use list-based arguments rather than shell strings, which prevents command injection. The operations are limited to standard development tools and are central to the skill's documented purpose.
  • [INDIRECT_PROMPT_INJECTION]: The project scaffolder interpolates user-provided metadata into project templates, including agent-specific context files like AGENTS.md. Ingestion points: Command-line arguments to new_project.py such as --name, --description, and --author. Boundary markers: Variable interpolation uses {{VAR}} delimiters within the templates. Capability inventory: The script performs file system writes and executes development tools. Sanitization: Basic slugification is applied to identifiers; prose fields are interpolated directly. Analysis: This represents a standard surface for scaffolding tools. The risk is minimal as the output consists of generated code and documentation intended for developer review.
  • [DYNAMIC_EXECUTION]: The skill generates Python source code and test files from predefined templates during the scaffolding process. Evidence: new_project.py creates project structures including __init__.py, main.py, and pytest files. Analysis: The generation follows simple templates and is a core feature of the scaffolding functionality, presenting no evidence of malicious code generation.
  • [EXTERNAL_DOWNLOADS]: The skill references the installation of an external architectural mapping tool. Evidence: SKILL.md and references/archify.md suggest using npx skills add tt-a1i/archify -g. Analysis: The reference is to an external utility for generating architecture diagrams. It is an optional, user-initiated action and does not involve hidden or automated downloads by the skill itself.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 3, 2026, 12:39 AM