What Creative Claw MCP server does
Creative Claw MCP server gives AI clients access to a hosted creative workspace for producing and managing AI media. The available capabilities span image generation and editing, video generation, speech, audio operations, asset storage, brand themes, reusable Characters, and multi-scene Film projects.
It supports workflows such as turning a still image into a video, creating product photos and ads, making UGC-style content, producing social posts, narrating media, and assembling short films. The service exposes a live model catalog, so clients can inspect available models and parameters instead of relying on a fixed list.
How it works
An MCP connection routes agent requests to Creative Claw’s hosted service. The agent can select or discover models, submit generation jobs, check video job status, and pass generated or uploaded media into later editing or campaign steps. One account can be used across the supported image, video, speech, music, and sound-effect models.
The tool set includes generation, asset management, brand configuration, Character management, Film project operations, cost estimates, credit balance access where supported by the client, and feedback submission. Pricing uses credits with pay-as-you-go billing; the README provides example credit costs for several workflows, but actual model availability and parameters should be checked at runtime.
Creative Claw MCP server is suited to agents that need more than a single generation call. For example, an agent can reuse a brand theme and Character, create several media components, add captions or voiceover, and assemble the result into a Film project.
Setup and configuration
ChatGPT users can install the Creative Claw plugin from the ChatGPT plugin directory, sign in, and issue requests in natural language. Cursor users can install it from the Cursor Marketplace; the first tool call opens browser authentication. Grok Bot uses the same plugin marketplace flow described in the README.
Hermes Agent can connect directly to the hosted endpoint with OAuth:
hermes mcp add creative-claw --url https://app.creativeclaw.co/mcp --auth oauth
hermes mcp login creative-claw
The README also documents a ChatGPT-specific MCP endpoint at https://app.creativeclaw.co/mcp/chatgpt. No user-provided model API keys are required according to the project documentation. Access instead depends on signing in to the Creative Claw account and using its credit-based pricing.
Tools and capabilities
The documented MCP tools include:
generate_image and load_image for image creation, editing, comparison, and loading.
generate_video and check_job for video jobs and status checks.
generate_speech, design_voice, clone_voice, transcribe, and isolate_audio for speech and audio workflows.
remove_background, upscale_media, trim_video, cut_and_reframe_video, scale_video, add_subtitles, extract_frames, and merge_media for media operations.
list_models, get_model_params, and search_examples for model and example discovery.
- Asset tools for searching, uploading, importing, updating, deleting, and confirming media uploads.
- Theme, Character, Film project, feedback, account, credit, and generation-estimate tools.
The documented model examples include Nano Banana 2, GPT Image 2.5, Seedream 5 Pro, Seedance 2.5, Gemini Omni, MiniMax H3 Max, ElevenLabs, and Cartesia. The live catalog exposed by list_models should be treated as authoritative.
Limitations and notes
The available model catalog, generation parameters, and reference limits can change, so clients should query the runtime tools rather than hard-code assumptions. Some account capabilities, including get_credits_balance, depend on whether the client exposes them. Voice cloning is described for a consenting speaker, and the service charges credits for many generation workflows even though cloning itself is described as free in the examples.