bmad-agent-ux-designer

Fail

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: HIGHCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill is configured to execute Python scripts located within the project's directory structure using uv run (specifically targeting {project-root}/_bmad/scripts/resolve_customization.py and resolve_config.py). This allows for arbitrary code execution if a project contains malicious scripts in these locations, as the agent automatically runs them during activation.
  • [DYNAMIC_EXECUTION]: The activation workflow includes executing arbitrary instruction strings provided in the activation_steps_prepend and activation_steps_append configuration arrays. This enables the execution of unvalidated instructions or commands defined in local or team-level customization files.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes a large amount of untrusted data from project files that can override agent behavior. \n
  • Ingestion points: Customization files (.toml), file-based persistent facts, and various task tickets (initiative, story, bug, spike Markdown files). \n
  • Boundary markers: No delimiters or explicit instructions to ignore embedded commands are used when loading this external content into the agent's context. \n
  • Capability inventory: The skill has access to script execution and tool installation capabilities, which can be abused if the ingested data contains malicious instructions. \n
  • Sanitization: There is no evidence of validation or sanitization for the ingested file contents before they are processed.
  • [EXTERNAL_DOWNLOADS]: The skill documentation suggests installing additional components via npx skills add. While the specific repository suggested belongs to the skill's author, the general mechanism involves downloading and installing external code into the execution environment.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Nov 13, 2026, 10:20 PM