Custom AI Providers
Custom AI providers let plugins register AI backends that appear as selectable providers inside Codebolt.
This section groups two related plugin patterns:
- Custom LLM Providers
- Custom Embedding Providers
They belong together because both are exposed through the provider-selection model and both are implemented through the plugin AI-provider surface.
Why This Is A Separate Section
These are not utility plugins. They extend Codebolt's model and inference backend layer.
Use custom AI providers when you need:
- a new chat/completion backend
- an in-house inference gateway
- OAuth-based provider login flows
- a specialized embedding backend
- custom routing or policy around AI models
Two Common Shapes
| Provider Type | What it adds |
|---|---|
| Custom LLM Provider | Chat/completion, streaming, tools, OAuth/login, model registration |
| Custom Embedding Provider | Vector generation for memory, search, and knowledge workflows |
Important Relationship
Embedding support is often bundled into an LLM provider plugin.
That means:
- some plugins expose both chat and embeddings
- some plugins are embedding-focused only
So the docs here are split for clarity, but the runtime surface is closely related.