Skip to content

AI assistants

Depicta works with any AI assistant or agent that can run CLI commands or make HTTP requests. Install a skill file or add the Depicta CLI, and your assistant gains the ability to generate, edit, and process images.

The fastest setup. Download a skill file from the dashboard — it includes your API key and full instructions for Claude.

  1. Sign up at depicta.ai/dashboard if you don’t have an account.

  2. Go to the Skills page in the dashboard sidebar.

  3. Choose your platform:

    • Claude Desktop — downloads a .skill ZIP package
    • Claude Code — downloads a .zip package
    • Raw Markdown — downloads the .md file directly
  4. Name your skill (optional) — e.g., “my-laptop” or “work-pc”. This helps you manage multiple devices.

  5. Click “Create & download” — save the file to your computer.

  6. Install the file:

    Double-click the .skill file. Claude Desktop imports it automatically.

  7. Test it — ask Claude: “Generate a photo of a sunset over the ocean.”

Cursor’s agent mode can run terminal commands, so it works with the Depicta CLI out of the box. To make it aware of the tool:

  1. Install the CLI and authenticate.
  2. Create .cursor/rules/depicta.mdc in your project with instructions about the depicta CLI — what commands are available and when to use them. You can use the Raw Markdown skill file from the dashboard as a starting point.

Cursor’s agent will then run depicta image "prompt" in the terminal when it needs to generate visuals. It also supports MCP servers if you prefer a richer integration.

Copilot’s agent mode (VS Code) can run terminal commands and supports custom instruction files:

  1. Install the CLI and authenticate.
  2. Create .github/copilot-instructions.md in your repo with instructions about the depicta CLI. You can base this on the Raw Markdown skill file from the dashboard.

You can also create a dedicated .github/agents/depicta.agent.md custom agent for image generation tasks. Copilot supports MCP servers as well.

Windsurf’s Cascade agent runs terminal commands and supports rules files:

  1. Install the CLI and authenticate.
  2. Create .windsurf/rules/depicta.md in your project with instructions about the depicta CLI. Use the Raw Markdown skill file as a starting point.

Cascade will run depicta commands in its dedicated terminal. MCP integration is also supported via ~/.codeium/windsurf/mcp_config.json.

Cline executes terminal commands with explicit user approval:

  1. Install the CLI and authenticate.
  2. Create .clinerules (or .clinerules/depicta.md) in your project root with instructions about the depicta CLI.

Cline also has first-class MCP support with a built-in marketplace.

Any AI assistant or agent that can run terminal commands can use Depicta via the CLI. The general pattern is:

  1. Install the depicta CLI and authenticate with your API key.
  2. Add an instruction/rules file to your project describing the CLI commands. Download the Raw Markdown format from the Skills page as a ready-made reference.

This works with Continue.dev (.continue/rules/), Amazon Q Developer (.amazonq/rules/), Augment Code (.augment/rules/), JetBrains AI / Junie, and any agent with terminal access.

The skill file teaches your AI assistant how to use Depicta. It contains:

  • Instructions for generating, editing, combining, and processing images
  • Available models and their capabilities
  • Guidelines search for better prompt writing
  • An embedded API key (unique to your download)

Your assistant reads the skill file and makes API or CLI calls on your behalf — or autonomously when it decides visuals are needed.

Downloading a new skill file for the same name revokes the previous key and creates a fresh one. Use this to rotate keys or set up a new device.

View and revoke active skill keys on the Skills page in the dashboard.