Skip to content
| Marketplace
Sign in
Visual Studio Code>Machine Learning>SovaraNew to Visual Studio Code? Get it now.
Sovara

Sovara

Preview

Sovara Labs

|
1 install
| (1) | Free
Debug LLM and agent workflows inside VS Code.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Sovara for VS Code

Sovara is a VS Code debugger for LLM and agent workflows.

Quick Start

With uv

uv add sovara
uv run so-record agent.py

With pip

pip install sovara
so-record agent.py

Then:

  1. Open your project in VS Code.
  2. Open the Sovara icon in the Activity Bar.
  3. Select a recorded run from the sidebar.
  4. Open the graph and inspect the run.

If sovara is installed in the selected Python environment, the extension can start so-server automatically. If you prefer to start it yourself:

Start the server manually with uv

uv run so-server start

Start the server manually with pip

so-server start

What You Can Do

  • Browse recent runs from the Sovara sidebar
  • Filter finished runs by name, label, tag, or code version
  • Open full graph tabs for individual runs
  • Click a node to inspect its input/output
  • Inspect attachments, code locations, labels, and applied priors
  • Open SovaraDB for project priors

Requirements

  • VS Code 1.74+
  • A Python environment with sovara installed
  • A project you run through so-record

Settings

  • sovara.pythonServerHost: Host for the local Sovara server
  • sovara.pythonServerPort: Port for the local Sovara server

Command

  • Sovara: Open Sidebar

Learn More

  • Website: https://sovara-labs.com

Support

  • Discord: Discord
  • Email: support@sovara-labs.com
  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft