Cursor
Cursor can send OpenAI-compatible chat and agent requests to LootVM when you supply a LootVM key and override the OpenAI base URL.
- Create a LootVM API key in the Console.
- Confirm model ids with
GET /v1/modelsfor that key. - Open Cursor Settings → Models → API Keys.
- Paste the LootVM key into the OpenAI API Key field and enable Use OpenAI API Key.
- Enable Override OpenAI Base URL and set:
https://api.lootvm.com/v1- Add or enable the model id exactly as returned by
/v1/models. - Select that model in Cursor Chat or Agent and send a short test prompt.
Tab autocomplete, Apply, Cloud/Background Agents, and Cursor’s proprietary models do not use this override. Turn the override off when you want Cursor-managed models that should not hit LootVM.
Verify outside Cursor
Section titled “Verify outside Cursor”export LOOTVM_KEY='your-lootvm-api-key'
curl --fail-with-body https://api.lootvm.com/v1/models \ -H "Authorization: Bearer ${LOOTVM_KEY}"If curl fails, fix the key or network path before debugging Cursor settings.
Troubleshooting
Section titled “Troubleshooting”| Symptom | Check |
|---|---|
| Invalid / unauthorized API key | Key pasted wrong, revoked, or not a LootVM-issued key |
Provider / resource not found (404) |
Base URL must end with /v1 and nothing after it |
| Model does not exist | Model id must match /v1/models for this key |
| Requests still hit OpenAI | Override OpenAI Base URL is off, or you selected a non-OpenAI Cursor-managed model |
| Cursor-managed models fail while override is on | Expected: the override routes OpenAI-format traffic to LootVM; disable override to use Cursor’s own models |
See Foundation (SDKs) for the shared OpenAI-compatible pattern and Chat Completions for the HTTP shape.