Zoo Code

Use the Zoo Code VS Code extension with the Nebul Inference API via the OpenAI Compatible provider.

Zoo Code is an open-source AI coding extension for VS Code and other VS Code-based editors. It reads and writes files, runs terminal commands, and coordinates work across modes. Because Nebul's Inference API is OpenAI-compatible, you can point Zoo Code straight at your Nebul project and skip third-party model providers.

Prerequisites

Visual Studio Code 1.84.0 or later (or another VS Code-based editor)
A Nebul AI Studio account with API access
An API key for your active project
A model ID from the Model Catalog

Installation

Install the extension from the VS Code Marketplace:

Open VS Code.
Go to Extensions (Ctrl+Shift+X / Cmd+Shift+X).
Search for Zoo Code and select the one published by ZooCodeOrg (marketplace page).
Click Install, then reload VS Code if prompted.

For VSCodium and other editors without Marketplace access, Zoo Code is also on the Open VSX Registry. Open the Zoo Code panel from the Activity Bar after installing.

Connect it to Nebul

Zoo Code talks to any OpenAI-compatible endpoint through its OpenAI Compatible provider:

Open the Zoo Code panel from the VS Code Activity Bar.
Click the settings icon to open the Zoo Code settings.
Set API Provider to OpenAI Compatible.
Fill in the fields:

Setting	Value
Base URL	`https://api.inference.nebul.io/v1`
API Key	Your Nebul API key from AI Studio
Model	A model ID from the Model Catalog, for example `zai-org/GLM-5.3`

Under Model Configuration, set the values for your model. Take the context window, max output tokens, and per-token prices from /v1/model/info so Zoo Code budgets context and spend correctly.
Save the settings.

The model must support tool calling

Zoo Code uses OpenAI-native tool calling exclusively, with no XML fallback. A model without tool calling cannot drive file edits or commands, so pick one with supports_function_calling in the catalog. zai-org/GLM-5.3 is a good default. See Function Calling & Tools.

Run it

Type a task into the Zoo Code panel and press Enter. Zoo Code proposes actions and waits for your approval. Usage appears under the API key's project in AI Studio.

Troubleshooting

Requests hit //models: the base URL ends with a slash. Enter https://api.inference.nebul.io/v1 with no trailing slash, so Zoo Code's appended paths stay single-slashed.
Requests hit /chat/completions instead of /v1/chat/completions: the base URL is missing the /v1 suffix. Zoo Code appends the endpoint path itself, so the /v1 has to be in the base URL.
401 responses: the API key is wrong, revoked, or from a different project. Create a fresh key in AI Studio if unsure.
Model not found: use the exact model ID from the Model Catalog, including the vendor prefix. Only models enabled for your project work.
Tool calls fail or the model is unusable: the model does not support OpenAI-native tool calling. Switch to one that does.

For provider-specific details, see Using OpenAI Compatible providers with Zoo Code.

Data and telemetry

Zoo Code sends anonymous usage data and error reports, linked to a per-install identifier, through PostHog. It does not collect your code or prompts. Turn it off in the Zoo Code settings by disabling Allow error and usage reporting.

Model requests go only to the provider you configure, which after this setup is Nebul. For how Nebul treats requests, see Privacy & Security.