Local AI Agent

Developed by Shravan Kumar SV

AI coding agent for VS Code with selectable Ollama and NVIDIA NIM providers. It supports local inference servers as well as NVIDIA's hosted API.

Powered by CorpRideConnect

It is designed to work with locally hosted AI models, allowing you to interact with your code through a VS Code chat interface while keeping your development workflow connected to your local AI infrastructure.

Features

• AI Coding Chat — Interact with your local AI coding agent directly from VS Code.
• AI Providers — Connect to Ollama, a self-hosted NVIDIA NIM, or NVIDIA's hosted API.
• Provider and Model Selection — Switch the active provider and model from the chat toolbar.
• Agent Selection — Choose the coding agent configuration from the chat interface.
• Scope Selection — Control the files or folders that the agent can work with.
• Current File Support — Work with the currently open file as the default file context.
• File and Folder Selection — Select the appropriate workspace files or folders from the chat interface.
• VS Code Sidebar Integration — Access the Local AI Agent directly from the VS Code sidebar.
• Settings — Configure the AI connection and related agent settings from the extension.
• Local Development Workflow — Designed for development environments where AI models can run on your own infrastructure.

Requirements

Visual Studio Code

The extension requires:

• Visual Studio Code 1.125.0 or later.

Ollama

Local AI Agent is designed to connect to an Ollama server running your selected AI model.

Install Ollama from:

https://ollama.com/

After installing Ollama, make sure the Ollama service is running and that the required model is available.

You can verify your installed models with:

ollama list

NVIDIA NIM

For a local NVIDIA NIM deployment, use the server address (for example, http://localhost:8000). An API key is optional unless your deployment requires one.

For NVIDIA's hosted API, use:

https://integrate.api.nvidia.com/v1

Create an NVIDIA API key, add an NVIDIA NIM provider in the extension Settings, and save the key there. The key is stored in VS Code SecretStorage rather than the extension's normal settings data. Use Test Connection to load agent-compatible model IDs and verify that the selected model can perform inference. NVIDIA's hosted model catalog also contains embeddings, rerankers, image-only pipelines, and parsers; those are intentionally excluded because they cannot act as a conversational coding agent.

NVIDIA Nemotron 3 Ultra's free hosted endpoint is supported with model ID:

nvidia/nemotron-3-ultra-550b-a55b

For this model, Local AI Agent automatically supplies NVIDIA's required coding-agent chat-template options for reasoning and tool calls. Hosted availability, quotas, and trial terms remain controlled by NVIDIA.

The provider supports NVIDIA's shared chat-completions endpoint and its model-specific hosted VLM endpoints. If a conversational model does not support OpenAI tool definitions, Local AI Agent automatically falls back to text-only chat for that model.

Getting Started

1. Install Local AI Agent in VS Code.
2. Open the Local AI Agent sidebar.
3. Open Settings.
4. Add and test an Ollama or NVIDIA NIM connection.
5. Select the provider and AI model from the Chat toolbar.
6. Select the appropriate agent and coding scope.
7. Open the Chat panel.
8. Start working with your code through the AI coding agent.

Chat Scope

The agent can work with the scope selected from the chat interface.

Depending on the selected scope, the agent can use:

• The currently open file
• Individual files
• Folders
• Workspace code

Selecting an appropriate scope helps keep the AI context focused on the code you want it to work with.

Privacy

Local AI Agent is designed for local AI development workflows.

When connected to a locally hosted Ollama or NVIDIA NIM instance, your code can remain within your local development environment. When using NVIDIA's hosted API, prompts and selected workspace context are sent to that service.

The actual privacy characteristics depend on how your provider and selected models are configured.

Configuration

Connection and agent configuration can be accessed through the Settings option in the Local AI Agent chat interface.

The available configuration options may change as the extension evolves.

Known Issues

This is an early release of Local AI Agent.

Some functionality may change as the extension continues to evolve. If you encounter an issue, please provide the relevant VS Code version, extension version, model information, and error details when reporting it.

Release Notes

0.0.1

Initial release of Local AI Agent.

• Added VS Code sidebar integration.
• Added AI coding chat.
• Added local AI model connection.
• Added model selection.
• Added agent selection.
• Added coding scope selection.
• Added Settings interface.
• Added file and folder context selection.

License

Local AI Agent is released under the MIT License.

About

Local AI Agent is an open-source VS Code extension developed by Shravan Kumar SV.

The project is powered by the development infrastructure and technology ecosystem of CorpRideConnect.

Visit CorpRideConnect