Skip to content
| Marketplace
Sign in
Visual Studio Code>Programming Languages>Atensor AINew to Visual Studio Code? Get it now.
Atensor AI

Atensor AI

Atensor

| (0) | Free
AI chat, agent mode, inline editing and completions, powered by Atensor.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Atensor AI

AI chat, agent mode, inline editing (Ctrl+K), completions and the no-code Build loop, powered by Atensor's models.

Works in Visual Studio Code, VSCodium and Atensor Code.

Getting started

  1. Open the Atensor tab in the secondary side bar, next to Chat (View → Appearance → Secondary Side Bar, or Ctrl+Alt+B).
  2. Sign in with your ATensorLab account.
  3. Ask in Chat, or select code and press Ctrl+K to edit it.
  4. To make a whole app from a description, click Build an App (the rocket) at the top of the Atensor tab; Back to Chat returns.

The agent reads your workspace freely and writes its edits straight away, so it can run and test them; every file it changed is listed under AI Changes, where you open the diff and Keep or Undo it. It asks in the chat before running a command, unless you choose Run commands automatically (risky ones, such as installing packages, deleting folders or pushing to git, still ask). The atensor.agent.* settings change this. Likely secrets are redacted before anything is sent.

MCP servers

The agent can use the tools of MCP servers. Atensor reads the configuration you may already have: your VS Code mcp.json, and a project's .vscode/mcp.json, .mcp.json or .cursor/mcp.json. Servers defined by a project run only after you allow them. Run Atensor: MCP Servers to see what's running, or Atensor: Open MCP Configuration to add one. Tools a server marks read-only run freely; the others ask first.

Models and limits

Atensor runs open models (such as Kimi K3, GLM 5.3 and DeepSeek V4 Pro); Auto picks the best one for each request, or choose one with the model button in Chat. Every request counts toward your ATensorLab account's daily messages, and an agent task makes one request per step.

Privacy

What you send (your messages, and the code Atensor reads to answer) goes to Atensor's servers and the model that answers, hosted on Baseten, which does not keep it or train on it. Files listed in .aiignore are never sent, and likely secrets are redacted first. Conversations are saved only on your computer. Your sign-in is an access key kept in your system's credential store; disconnect any editor on the Account page at chat.atensorlab.com.

Requirements

Visual Studio Code 1.136 or later (or VSCodium / Atensor Code), and a free ATensorLab account.

License

Proprietary. © 2026 ATensorLab. All rights reserved; see the License link on this page for the full terms.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
  • Your Privacy Choices
  • Consumer Health Privacy
© 2026 Microsoft