Frontend Integrations
CoreCube exposes an OpenAI-compatible /v1/ API, which means any chat frontend that supports a custom OpenAI-compatible endpoint can use CoreCube as its knowledge backend.
How it works
When a user sends a message in their chat frontend, the frontend calls CoreCube's /v1/chat/completions endpoint. CoreCube searches its knowledge base, injects the relevant context into the prompt, and returns a cited response — all transparently, using the same format the frontend already understands.
Per-user access control
CoreCube supports compartment and sensitivity-level access control per user. To enforce this in a multi-user frontend, you need a service API key — a key that identifies the calling application (e.g. OpenWebUI) and resolves the end user's identity from a header or request field on each request.
Each frontend has its own mechanism for forwarding the logged-in user's identity. The guides below cover the exact setup for each supported frontend.
Supported frontends
| Frontend | Per-user access control | Integration method | Guide |
|---|---|---|---|
| OpenWebUI | ✅ Full | Header X-OpenWebUI-User-Email | Setup → |
| LibreChat | ✅ Full | Header X-Cube-User (configured in librechat.yaml) | Setup → |
| Onyx | ✅ Full | Request body user field | Setup → |
| AnythingLLM | ⚠️ Shared user only | Static header | Setup → |
:::info Not seeing your frontend? CoreCube accepts any OpenAI-compatible client using a personal API key — one key, one user, no extra configuration. If your frontend doesn't support per-user identity forwarding, a personal key is the right choice. See API key types. :::