bmad-loop-setup
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Downloads and installs the
bmad-looporchestrator tool directly from the vendor's Git repository (github.com/bmad-code-org/bmad-loop.git) usinguv tool install. This action is central to the skill's function and targets the author's official source. - [COMMAND_EXECUTION]: Executes several CLI commands to manage the environment and project state, including
uv tool list,bmad-loop init,bmad-loop validate, and a local configuration scriptresolve_config.pyfound within the project's_bmad/scripts/directory. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external tool outputs and configuration files, creating a potential surface for instruction injection.
- Ingestion points: Processes output from
uv tool list,bmad-loop --version, andbmad-loop validate, as well as metadata fromassets/module.yaml. - Boundary markers: No specific delimiters or safety instructions are used when interpolating these data points into the agent's workflow.
- Capability inventory: Possesses capabilities to write files, install Python packages, and execute shell commands.
- Sanitization: No explicit sanitization or validation of the ingested tool output or YAML content is observed.
Audit Metadata