Skip to content

Integrations

Once you have validated a prompt or workflow in the AI Playground, you can bring Picsart into the AI agent or coding assistant you already work in. Generate images, animate a video, remove a background, or synthesize audio without switching windows.

These guides cover the exact setup for each agent.

Supported agents

AgentGuide
Claude CodeClaude Code
CursorCursor
WindsurfWindsurf
ChatGPTChatGPT
Codex (OpenAI)Codex
VS Code CopilotVS Code

Skills vs MCP: which to connect

Both connect Picsart to an agent, but they work differently.

Skills are pre-built instruction bundles. Install a skill and the agent already knows how to use Picsart — you just describe the task in plain English. Fastest to set up. Best when you want to generate from a conversation without managing tool schemas.

MCP exposes every catalog tool directly: picsart_generate, picsart_preflight, picsart_remove_bg, and more. Use MCP when you want the agent to inspect cost before generating, validate parameters, or chain multiple operations in one turn.

You can use both at the same time — they do not conflict.

SkillsMCP
InstallOne command or ZIPOne config block
Agent knows PicsartYes, pre-builtVia tool schema only
Can quote cost before generatingNoYes
Can chain toolsNoYes
Best forConversational generationWorkflow automation inside the agent

Prerequisites for all integrations

  1. Install the gen-ai CLI — see Installation.
  2. Authenticate once: gen-ai login (browser OAuth).

Skills and MCP both drive the CLI internally. The session from step 2 covers all surfaces — no separate credential per agent.

Not using an agent?

  • Want to experiment first? Use the AI Playground — no install required.
  • Need scheduled or batch generation? Use the CLI.
  • Building a product? Use the API.

Built on @picsart/ai-sdk · gen-ai CLI · Picsart MCP · Skills