setup-ml-project
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local commands using
python -m skore_skillsto detect environment configurations and persist user policies. These operations are restricted to project state management and status reporting. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local workspace status checks (
status.skills,status.setup.pending) to determine its logical flow. While this constitutes an external data ingestion surface, the source is the local environment via internal tooling, and the skill implements strict boundary logic to ensure user-selected pieces are the only ones processed. - [REMOTE_CODE_EXECUTION]: The instructions contain an explicit security constraint prohibiting the use of
curl | shor similar untrusted remote execution patterns during environment manager setup. - [DYNAMIC_EXECUTION]: The skill utilizes a composition pattern to load and execute sub-skills (e.g.,
setup-git,setup-workspace). This is a standard platform mechanism for modular functionality and is used here in a controlled sequence based on user input.
Audit Metadata