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

Sovara

Preview

Sovara Labs

| (0) | Free
Inspect, edit, and replay agent workflows without leaving 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

Inspect, edit, and replay agent workflows without leaving VS Code.

Sovara turns recorded runs into an interactive graph inside the editor so you can trace prompts, outputs, dependencies, and cached reruns in one place.

Features

  • Browse recent runs from the Sovara sidebar
  • Open full graph tabs for individual runs
  • Inspect inputs, outputs, attachments, and code locations
  • Edit a node and rerun only downstream work
  • Reconnect to the local Sovara server automatically

Requirements

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

Quick Start

  1. Install sovara in the Python environment VS Code should use:

    uv add --dev sovara
    # or
    pip install sovara
    
  2. Open your project in VS Code.

  3. Record a run:

    so-record python your_script.py
    
  4. Open the Sovara icon in the Activity Bar and select a run.

By default the extension connects to 127.0.0.1:5959. It can start so-server automatically when sovara is installed in the selected Python environment. If you prefer to start it yourself:

so-server start

Settings

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

Command

  • Sovara: Open Sidebar

Support

  • Repository: https://github.com/SovaraLabs/sovara
  • Issues: https://github.com/SovaraLabs/sovara/issues
  • Email: hello@sovara-labs.com
  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft