Python MCP Server for Gemini Image Model Updated for Kiro CLI 3 and MCP SDK 2.x

A Python MCP server powering Google's Gemini Nano Banana 2 Lite image model has been updated to work with Kiro CLI 3, MCP SDK 2.x, and the revised Gemini Interactions API. Key changes include replacing FastMCP with MCPServer and upgrading google-genai from version 1.x to 2.x, which became necessary after Google removed the legacy API schema on June 8, 2026. The updated server is registered within Kiro with explicit permission rules and validated end-to-end through a headless Kiro 3 session. A new unit test was added that builds a real steps-schema Interaction object, ensuring dependency mismatches cause loud test failures rather than silent API errors. The core server logic in server.py remains unchanged, with differences limited to how Kiro launches, permissions, and locates the skill.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in