Clipia

Generate images, video, speech, music and presentations with 60+ models, including Veo, Kling and Seedance.

Documentation

Connect

Pick your agent — IDEs and CLIs need an API key from the Developer Console, while claude.ai and ChatGPT connect by signing in to your Clipia account (OAuth), no key required.

Terminal · CLI

Run a single command in your terminal (key from the Clipia Developer Console): One command claude mcp add --transport http clipia https://mcp.clipia.ai/mcp --header "Authorization: Bearer clipia_YOUR_KEY"
Done. Ask Claude: “Generate a neon city image with Clipia” — the preview lands right in the terminal.

Create a key in the Clipia Developer Console. Keys with the clipia_test_ prefix run in a sandbox without spending credits.

Also listed in:MCP Registry Smithery Glama PulseMCP Cursor Directory

10 core tools for AI agents

Image, video, speech and music generation, status, catalog and prompt tools. Five additional agent tools appear when their capabilities are enabled.

generate_image

Text-to-image or reference-based editing. Usually returns the finished image in a single call — with a preview.

generate_video

Video from text or a start frame (image-to-video). Returns request_id and the credit cost immediately.

generate_audio

Text-to-speech: choose a voice and language and receive an MP3. Use it on its own or as a voiceover in the video pipeline.

generate_music

Background music and soundtracks from a mood, genre and tempo description. Produces an instrumental MP3 by default.

wait_generation

Wait for completion: long-poll until a terminal status.

get_generation

Instant status and result: webp preview + full-quality original_url.

list_models

Model catalog with types, capabilities and credit prices.

get_model

Parameters of a specific model (input_schema) and its price.

get_balance

Credit balance and 30-day usage of the key.

search_templates

Search 3,500+ curated prompts exposed to MCP from the wider template library (hybrid, RU/EN).

Read the complete AI-agent and tool reference

A sandbox for integrations

Keys with the clipia_test_ prefix return instant mock results without spending credits — handy for debugging agents and CI. Live keys include the exact cost of every generation in the response, and get_balance shows the remaining credits.

FAQ

Model Context Protocol is an open standard that lets AI agents (Claude, ChatGPT, Cursor and others) plug in external tools. Clipia MCP always exposes 10 core agent tools for images, video, speech, music, status, catalog access and prompt search. Five capability-dependent tools add chat, scenario planning, video composition and presentation generation or editing. Seven app-only helpers power interactive cards and are not exposed to the AI agent.

For Claude Code, Cursor, VS Code, Windsurf, Codex and Gemini CLI — yes, a key from the Clipia Developer Console. claude.ai and ChatGPT need no key: they connect by signing in to your Clipia account (OAuth).

Exactly like on the website: from your Clipia account balance. Every tool response includes the precise credit cost of the generation, and get_balance shows the remainder. No hidden MCP markup.

Every active model in the Clipia catalog — 60+ for images and video. The agent fetches the list via list_models — up to 100 models per call (limit: 100), the next page via offset — including prices and capabilities; you can request a specific model by name.

Yes. In claude.ai (web, desktop, mobile) the result renders as an interactive card with live progress and an “Original” button. In Claude Code the preview lands right in the terminal. Other clients receive full-quality file links.

Images usually take up to 30 seconds and often arrive in a single call. Video renders in 1–10 minutes: the agent tracks the status via wait_generation and continues as soon as it’s ready.

Keys are stored as hashes only, permissions are scoped, and a key can be revoked in one click in the Developer Console. OAuth connections use PKCE and one-time codes; access is strictly limited to your own account.

Yes. The MCP server runs on top of the same public API: idempotent retries without double charges, a queue, webhooks and SDKs. For pipelines there is the clipia_test_ sandbox and a REST API documented on the developers page.