Tool setup
Windsurf Integration Guide
If your Windsurf version supports custom providers, you can connect SeaLink as an OpenAI-compatible endpoint. Menus and capabilities may vary slightly between versions.
Prerequisites
- Windsurf version ≥ 1.0 (click profile icon → About to check)
- A valid SeaLink API Key (starts with sk-sealink-)
- Network access to sealink.io (port 443, HTTPS)
Setup Steps
Open Windsurf Settings
Click the gear icon in the bottom-left corner and select 'Settings'. You can also use the shortcut: Ctrl+, (Windows/Linux) or Cmd+, (Mac) to open settings directly.
Check for custom provider support
Search for 'provider' or 'model' in settings. You can only connect SeaLink directly if your Windsurf version shows a 'Custom' or 'OpenAI Compatible' provider option. If your version doesn't support this, try Cursor, Cline, or Claude Code instead — they all offer standard custom API configuration.
Add a custom provider
Once confirmed, fill in: API Base URL: https://test.sealink.io/v1 API Key: Your SeaLink key (starts with sk-sealink-) Don't append /chat/completions or any other path.
Select a model
After saving, select or type a SeaLink model ID in Cascade or the chat panel. We recommend starting with gpt-4o-mini to quickly verify the connection, then switching to the model you need for actual coding work.
Verify
Send a short test message in chat or Cascade. Keep in mind that different Windsurf versions may not route every editor feature through the custom provider — this depends on Windsurf's implementation.
Operating System Notes
Windows
Settings file: %APPDATA%/Windsurf/User/settings.json. If settings don't stick through the UI, you can edit this JSON file directly.
macOS
Settings file: ~/Library/Application Support/Windsurf/User/settings.json. Use Cmd instead of Ctrl for all shortcuts.
Linux
Settings file: ~/.config/Windsurf/User/settings.json. Same path whether installed via AppImage, snap, or deb.
Common Issues
No Custom / OpenAI Compatible provider option
Your Windsurf version may not expose a custom OpenAI-compatible provider. If so, you can't connect SeaLink directly through the GUI. Try Cursor, Cline, or Claude Code instead — all three offer straightforward custom API configuration.
SeaLink models don't appear in Cascade dropdown
First, double-check that your Base URL ends with /v1. If the URL is correct, verify your API Key is active. Windsurf's model picker may filter the list — try typing gpt-4o-mini manually as a workaround.
401 or authentication failed
Your API Key is likely incorrect or expired. Go to your SeaLink dashboard, confirm the key is still active, then re-copy and paste it — watch out for leading or trailing spaces.
Garbled or malformed model output
Some models may need extra parameter tuning in Windsurf. Start by verifying basic connectivity with a general-purpose chat model (like gpt-4o-mini). Once that works, switch to a coding-specific model from the catalog.