Tool integrations
Most tools need three settings: API address, Fenno API key, and model ID. Choose a guide for your tool before configuring it: addresses are not interchangeable between protocols.
Choose your workflow
Connect Claude Code, Codex or OpenCode to a provider and start in your project directory.
Configure Cline or Cursor and check which editor features actually use your API key.
Add a provider in Cherry Studio or Chatbox, enable models and start a conversation.
Use the OpenAI Python or JavaScript SDK for regular requests and streaming output.
Before you start
- Create an API key in the Fenno console and check your available quota.
- Copy an exact model ID from the console or model list. Do not use a display name or invent a provider prefix.
- Check protocol and capability support. Codex needs Responses; coding agents also need streaming and tool calling.
- Back up existing configuration. Use a new conversation and a simple prompt before allowing changes to an important project.
A tool subscription and Fenno API usage are billed separately. Bringing your own key does not unlock every paid client feature or make requests free.
Tools and protocols
| Tool | Connection used in this guide | Documentation |
|---|---|---|
| CC Switch | Manage Claude Code / Codex providers | Setup |
| DeepSeek Harness | Custom model provider | Setup |
| WorkBuddy | Custom Chat Completions model | Setup |
| Claude Code | Anthropic Messages | Setup |
| Codex | OpenAI Responses | Setup |
| OpenCode | Custom OpenAI-compatible provider | Setup |
| Cline | OpenAI Compatible | Setup |
| Cursor | Custom OpenAI endpoint, subject to client restrictions | Compatibility and setup |
| Cherry Studio | Custom OpenAI provider | Setup |
| Chatbox | Custom OpenAI-compatible provider | Setup |
| OpenAI SDK | Server-side Python / JavaScript | Migration |
| Gemini CLI | Native Gemini API-key authentication | Setup |
| Zed | Custom OpenAI-compatible provider | Setup |
| LangChain | ChatOpenAI with a custom endpoint | Setup |
| LlamaIndex | OpenAILike and a separate embedding adapter | Setup |
| Other tools | Match the host, base URL and request path to the protocol | General setup |
Verify the actual connection
Send “Reply with OK only”, then match the request time, model and usage in the Fenno console. A model claiming to use Fenno is not proof of routing.
Test streaming, image input and tool calling separately after the text request works. A working connection does not imply support for every capability. See the general checklist and errors and retries.