Tool setup
Cursor Integration Guide
Add SeaLink to Cursor and use your enabled chat and coding models through an OpenAI-compatible endpoint. Note: some built-in Cursor features (like tab completion) may still use Cursor's own models.
Prerequisites
- Cursor version ≥ 0.40 (click your avatar → About to check)
- A valid SeaLink API Key (starts with sk-sealink-)
- Network access to sealink.io (port 443, HTTPS)
Setup Steps
Open Cursor Settings
Open Cursor, click the gear icon in the top-right corner, then select 'Models'. You can also use the keyboard shortcut: Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (Mac), then type 'Cursor Settings' to jump straight there.
Find the OpenAI API Key section
Scroll down in the Models page until you see 'OpenAI API Key'. If it's not there, make sure your Cursor is version 0.40 or newer (older versions may not expose this option). Once you find it, toggle on 'Override OpenAI Base URL'.
Enter your SeaLink base URL
In the 'Base URL' field, enter: https://test.sealink.io/v1 Important: don't skip the /v1 suffix — using just the domain name will result in an empty model list.
Enter your API Key
Paste your SeaLink API Key (starts with sk-sealink-) into the 'API Key' field. Don't have a key yet? Head to your SeaLink dashboard → Keys, create one, and copy the full key string.
Pick a model
After saving, select or type a SeaLink model ID in Cursor's model picker. We recommend starting with gpt-4o-mini to verify connectivity, then switching to the model you need for your actual work.
Verify it works
Send a short message in the Chat panel to confirm everything is connected. One thing to keep in mind: Cursor's custom API key mainly covers standard chat models. Tab completion, some inline edits, and other specialized features may continue using Cursor's built-in models — that's a Cursor design choice, not a SeaLink limitation.
Operating System Notes
Windows
Settings file: %APPDATA%/Cursor/User/settings.json. If pasting the Base URL doesn't seem to work, try right-clicking and selecting 'Paste as plain text'.
macOS
Settings file: ~/Library/Application Support/Cursor/User/settings.json. Use Cmd instead of Ctrl for all shortcuts.
Linux
Settings file: ~/.config/Cursor/User/settings.json. Same path for both AppImage and snap installations.
Common Issues
SeaLink models don't appear in the model list
This usually means the Base URL is incomplete. Double-check that it ends with /v1 — if you only entered the domain name without /v1, Cursor can't fetch the model list. Add /v1 and refresh the page.
401 Unauthorized error
Your API Key isn't being accepted. Go to your SeaLink dashboard, make sure the key is still active, then re-copy and paste it — watch out for any extra spaces before or after the key.
Slow responses or timeouts
Try switching to a lightweight model (like gpt-4o-mini) to test basic connectivity first. If that works but a specific model is slow, it may be under heavy load. If all models time out, check that your network can reach sealink.io and look at the status page for any ongoing issues.
Some models don't show up in Cursor
Cursor's model picker applies its own filtering and may not display every available model. Try typing the model ID manually. If that still doesn't work, test with another standard OpenAI-compatible chat model to narrow down the cause.
Tab completion does not use SeaLink
This is a known limitation of Cursor's custom API key feature. Custom keys primarily cover standard chat conversations, while tab completion and some inline editing features continue to use Cursor's own models. This is how Cursor is designed — it's not specific to SeaLink.