MCP Integration
PicRenew ships a Model Context Protocol (MCP) server so you can use photo restoration directly inside AI clients like Claude Desktop, Cursor, and other MCP-compatible tools — without writing any code. Model Context Protocol (MCP).
Endpoint
The MCP server uses Streamable HTTP transport (stateless). The same Bearer API key used for the REST API authenticates MCP requests. See the Authentication page for key format and security best practices.
Configure Claude Desktop
Add PicRenew to your Claude Desktop MCP configuration file ( ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
Restart Claude Desktop after saving. You should see the PicRenew tools available in the tool picker.
Configure Cursor
Add the following to your Cursor settings.json (or the MCP panel in Cursor settings):
Available tools
restore_photo
Restores and enhances a photo from a public URL.
| Parameter | Type | Required | Description |
|---|---|---|---|
image_url | string | Yes | Public HTTPS URL of the photo to restore |
preset | string | No | standard | enhance | colorize | upscale (default: standard) |
resolution | string | No | 1k | 2k | 4k (default: 1k) |
remove_frame | boolean | No | Remove photo border before restoring (default: false) |
colorize | boolean | No | Apply AI colorization (default: false) |
callback_url | string | No | HTTPS webhook URL to receive the completed result |
Returns a JSON object with id, status, Use the id to poll with get_restoration.
get_restoration
Returns the current status of a restoration job.
| Parameter | Type | Required | Description |
|---|---|---|---|
id | string (UUID) | Yes | Restoration ID returned by restore_photo |
Returns a JSON object with status, output_url. Poll this tool until status or complete or failed.
Example conversation
Credits and billing
MCP requests consume credits from your PicRenew account, identical to the REST API. Credits are debited when the job is created. View your balance and usage on the account page or billing page.
Notes
- The MCP server is stateless — no session is maintained between requests. Each tool call is independent.
- Rate limits are shared with the REST API (60 requests/minute per API key).
- Images are fetched from the provided URL and re-uploaded to PicRenew storage before processing begins.