kie-ai

This MCP server exposes Kie.ai's Nano Banana image generation and editing capabilities along with Veo3 video generation. It also includes a SQLite-backed task tracker and smart endpoint routing to poll task status. Tools available include generate_nano_banana for text-to-image generation, edit_nano_banana for image editing using natural-language prompts (up to 5 input images), and generate_veo3_video for text-to-video generation with optional image inputs and model variants. You can query task status with get_task_status, list recent tasks with list_tasks, and retrieve 1080p upgrades for Veo3 outputs with get_veo3_1080p_video. All endpoints are wired to Kie.ai backend APIs and handle common errors such as unauthorized API keys, rate limits, and validation problems.

Prerequisites:\n- Node.js 18+ (and npm)\n- Access to a Kie.ai API key\n\nInstallation steps:\n1) Install the MCP server globally from NPM:\nbash\nnpm install -g @andrewlwn77/kie-ai-mcp-server\n\n2) or install from source:\nbash\n# Clone the repository\ngit clone https://github.com/andrewlwn77/kie-ai-mcp-server.git\nc d kie-ai-mcp-server\n\n# Install dependencies\nnpm install\n\n# Build the project (if applicable)\nnpm run build\n\n3) Set up configuration and environment variables:\n- Obtain a Kie.ai API key from https://kie.ai/api-key\n- Set KIE_AI_API_KEY in your environment (and optionally KIE_AI_BASE_URL, KIE_AI_TIMEOUT, KIE_AI_DB_PATH)\n\n4) Run the server (example using node with built dist index):\nbash\nexport KIE_AI_API_KEY="your-api-key-here"\nnode dist/index.js\n

Tips:\n- Store the API key securely; do not commit it to version control.\n- Use KIE_AI_DB_PATH to point SQLite to a persistent location for task tracking.\n- If you deploy in a container or MCP client, ensure the environment variables are passed to the process.\n- For production, consider enabling a persistent DB path and setting appropriate timeouts.\n- The MCP server exposes multiple tools; use list_tasks to monitor progress and handle retries for failed tasks.\n- If you encounter 'Unauthorized' errors, re-check the KIE_AI_API_KEY validity and permissions at https://kie.ai/api-key.