magicpath
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill uses the
!command`` syntax inSKILL.mdto executenpx -y magicpath-ai infowhen the skill is loaded. This command retrieves the user's authentication status and active project context. This is a legitimate use of project-specific tooling.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external UI component source code and metadata from the MagicPath platform viainspectandsearchcommands.\n - Ingestion points: Component source code is retrieved via
npx -y magicpath-ai inspectandnpx -y magicpath-ai code context.\n - Boundary markers: None explicitly defined in the instructions for the ingested code content.\n
- Capability inventory: The skill has capabilities to execute shell commands via the
magicpath-aiCLI, clone Git repositories, and write files to the local project directory.\n - Sanitization: There are no documented steps for sanitizing or escaping instructions that might be embedded in the component code or metadata before the agent processes them.\n- [COMMAND_EXECUTION]: The skill uses the
npx -y magicpath-aicommand to interact with the vendor's API, which downloads and executes the package from the npm registry.
Audit Metadata