GetRun MCP Server
Remote Model Context Protocol server for
GetRun — generate images, video, audio and chat with
61 AI models from 16 providers from inside any MCP client.
No API key to paste. The server speaks OAuth 2.1 with dynamic client
registration, so a client is authorised once through a consent screen and
billed against the same Rupiah balance as the web app and the REST API.
This repository holds the connector's public metadata and setup guide. The
service itself is hosted — there is nothing to install and no server to run.
Install
Claude (claude.ai)
Settings → Connectors → Add custom connector. Name it GetRun, paste the
URL above, save, then press Connect and approve the permissions.
Claude Code
Then run /mcp inside a session, pick getrun, and complete the browser
consent flow once.
Cursor
~/.cursor/mcp.json:
Any other MCP client
Point it at the URL. Clients that only speak stdio can bridge with
mcp-remote:
Tools
Catalogue
61 models from 16 providers, as of September 2026:
The live list is always at GET /v1/models —
this table is a snapshot, that endpoint is the source of truth.
Billing
Identical to the REST API: every tool calls /v1 in the same process, so
pricing, balance reservation, schema validation and rate limits are the same,
and results appear in the dashboard's usage history.
Generation tools spend real balance. getrun_estimate_price returns the cost
of a call before you make it, and a task that fails is never charged.
Also available
Registry metadata
server.json is the manifest published to the
official MCP Registry under
the DNS-verified namespace id.getrun.
Support
[email protected] — include the task_id or request_id when reporting a
failed generation.
Operated by PT Core Digital Asia (GetCore), Indonesia.