developing-genkit-dart

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs users to install the Genkit CLI using a shell script from cli.genkit.dev and the Node.js package manager (npm install -g genkit-cli). These are official installation methods for the Genkit framework. Other external dependencies include Dart packages like schemantic and various Genkit plugins.
  • [COMMAND_EXECUTION]: Documentation includes commands for running the Genkit developer UI (genkit start), analyzing code (dart analyze), and executing Model Context Protocol (MCP) servers via npx. These commands are intended for a local development environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill describes an attack surface through its filesystem and MCP plugins, which allow agents to interact with local files.
  • Ingestion points: Filesystem middleware (references/genkit_middleware.md) and MCP Host (references/genkit_mcp.md) allow the agent to read local files (e.g., read_file, list_files).
  • Boundary markers: The skill documentation highlights the ToolApprovalMiddleware in references/genkit_middleware.md, which provides a mechanism to intercept and require explicit user approval for sensitive tool executions.
  • Capability inventory: Capabilities include read_file, write_file, and search_and_replace within restricted directories, as well as network requests to configured AI model providers.
  • Sanitization: No specific data sanitization logic is shown in the provided code snippets, relying instead on the framework's internal handling and manual tool approval gates.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 11:02 PM