Use CoralBricks with Cursor

Route Cursor's chat and agent inference to CoralBricks open models — GLM and Kimi with up to 1M context — with a guided one-click setup.

Cursor supports custom OpenAI-compatible endpoints, so its Chat and Agent modes can run on CoralBricks inference. Our companion extension turns the setup into a couple of clicks — and keeps it working afterwards.

Requires a paid Cursor plan (Pro or higher). Cursor's Free plan locks model selection to "Auto" and cannot select custom models — that's a Cursor platform restriction. On a free plan, use OpenCode instead.

1. Install the CoralBricks extension

curl -fsSLO https://www.coralbricks.ai/downloads/coralbricks-cursor-0.2.1.vsix
cursor --install-extension coralbricks-cursor-0.2.1.vsix

Restart Cursor (or run "Reload Window") after installing.

2. Connect

Connect Cursor

The button opens Cursor and starts the setup wizard. Have your API key (ak_...) ready — the wizard validates it live against the gateway, then walks you through the three pastes Cursor requires (its model settings are paste-only by design):

Base URL — copied to your clipboard; paste into "Override OpenAI Base URL" under the API Keys section of the Models tab (the wizard opens the right tab for you)
API key — copied next (auto-clears from the clipboard after 60s); paste into the "OpenAI API Key" field and enable the toggle
Model slugs — copied one per click; add via "+ Add model"

The wizard's verify step then probes the gateway and confirms the toggle is on. You can re-run it any time: Command Palette → CoralBricks: Connect.

Available models

Model	Slug	Context	Input $/M	Output $/M
GLM 5.2	`glm-5.2-fp4`	1M	$1.40	$4.40
Kimi K3	`kimi-k3`	1M	$3.00	$15.00
Kimi K2.6	`kimi-k2.6`	256K	$0.76	$4.00
GPT-OSS 120B	`gpt-oss-120b`	128K	$0.15	$0.60

Cached input tokens are always free.

What routes through CoralBricks

Chat and Agent (Cmd/Ctrl+L) requests for the models you add — including streaming tool calls in agent mode. Tab autocomplete and inline edit (Cmd/Ctrl+K) always use Cursor's own backend; no third-party provider can change that.

The toggle guardian

Cursor has a long-standing bug where the custom "OpenAI API Key" toggle turns itself off. The extension watches for it (read-only) and re-enables the toggle through Cursor's own command the moment it happens, with a notification — so your Coral routing stays on without babysitting.

Manual setup (no extension)

Cursor Settings → Models → expand API Keys:

"Override OpenAI Base URL" → https://inference.coralbricks.ai/v1
"OpenAI API Key" → your ak_... key; enable the toggle
"+ Add model" → add the slugs from the table above
Pick the model in the chat picker

Notes

The API is OpenAI wire-format at https://inference.coralbricks.ai/v1 — the same key also works in OpenCode, Cline, Continue, and anything else that speaks OpenAI-compatible providers. See the API reference for the full surface.
Your key is stored in the editor's OS keychain (SecretStorage), never in settings files. The extension talks only to the CoralBricks gateway.