Kieai integration for Claude and Codex

Add AI-generated instrumental accompaniment to an existing vocal track. Use when you need to generate instrumental music to accompany a vocal recording, allowing customization of style, mood, and creative parameters. The task completion will be notified via the callback URL.

Add AI-generated vocals to an existing instrumental music track. Use this action when you want to add singing or vocal tracks to accompany an existing instrumental audio file. The action submits a vocal generation task and returns a task ID that can be used to track the generation status. This action is asynchronous - use the returned taskId to poll for completion via the callback URL or check task status separately.

Create a task for unified model access through the Market API. Submit tasks to any supported AI model (Kling, Sora2, Hailuo, Wan, Google, Ideogram, ElevenLabs, etc.) through a single unified interface. Specify the model and its input parameters, then receive a task ID to poll for completion. Use this action when you need to generate content (text-to-speech, images, video, etc.) via Kie.ai's unified Market API with access to multiple AI providers through one endpoint.

Create a music video (MP4) combining generated music with visuals. Use this action when you need to generate an MP4 music video from previously generated music tracks. The action requires a task ID from the music generation process and an audio ID for the specific track. The video generation is async - results are delivered to the specified callback URL. Once the video is ready, the callback URL receives a POST request with: - code: 0 for success, 500 for error - data.task_id: The unique task identifier - data.video_url: The video URL (valid for 14 days)

Access GPT Codex capabilities through the KIE.AI Market API. Use when you need to leverage advanced code understanding and generation abilities via the Codex model for tasks such as code completion, refactoring, bug explanation, or general programming assistance. The action supports both single-turn queries (plain string input) and multi-turn conversations (list of message objects with roles). Optional features include configurable reasoning depth, web search for up-to-date information, and function calling for structured tool use. Note: Web search and function tools are mutually exclusive and cannot be used together in the same request. When streaming is enabled (stream=True), the response is delivered incrementally via SSE events.