Enhanced Symbol Search Across Installed Packages
Navigating unfamiliar codebases or exploring third-party libraries often requires jumping between your editor and external documentation. The March 2026 release of the Python extension for Visual Studio Code addresses this by extending Pylance's workspace symbol search to include symbols from packages installed in your active virtual environment. Previously limited to your workspace files, this feature now lets you quickly locate functions, classes, and other symbols defined in site-packages without leaving VS Code.
How It Works
When enabled, the Python › Analysis: Include Venv In Workspace Symbols setting instructs Pylance to index symbols from the active virtual environment's site-packages directory. This means a quick Cmd/Ctrl+T can surface definitions from libraries like NumPy or Flask as if they were part of your project. To keep results focused, the indexer only includes symbols exported via __init__.py or __all__ for packages without py.typed annotations. You can further fine-tune indexing depth per package using the Python › Analysis: Package Index Depths setting, controlling how many sub-module levels Pylance searches.
Enabling the Feature
Because indexing installed packages can impact performance, this is an opt-in feature. To try it:
- Open Settings (Cmd+, on macOS or Ctrl+, on Windows/Linux)
- Search for "Include Venv In Workspace Symbols"
- Check the box under Python › Analysis
Once enabled, you'll immediately notice richer code exploration when working with third-party dependencies, while the default experience remains unchanged for others.
Experimental Rust-Based Parallel Indexer
Behind the scenes, Pylance's indexing engine—responsible for completions, auto-imports, and workspace symbol search—receives a major experimental upgrade. The March 2026 release introduces a new Rust-based parallel indexer that runs out-of-process, replacing the previous single-threaded implementation. In internal tests, this indexer achieves an average 10× speed improvement on large Python projects, resulting in faster completions after workspace open and a more responsive IntelliSense experience.
Performance Gains
The new indexer parses and indexes code in parallel, leveraging Rust's efficiency to handle complex project structures without blocking the editor. Early benchmarks show significant reductions in indexing time for projects with hundreds of files and deep dependency trees. Small projects may see minimal difference, but for large codebases, the improvement can be transformative.
How to Enable
This is intentionally experimental; the team wants to validate reliability across diverse environments before making it the default. To activate it:
- Open Settings (Cmd+, / Ctrl+,)
- Search for "Parallel Indexing"
- Check Enable Parallel Indexing (Experimental) under Python › Analysis
Alternatively, add the following to your settings.json:
"python.analysis.enableParallelIndexing": true
After enabling, reload VS Code (Cmd/Ctrl+Shift+P → Reload Window) to start the new indexer cleanly. This setting has the most impact on larger projects.
We Want Your Feedback
If you try the new indexer, share your experience. Whether you observe faster completions, slower behavior, or unexpected issues, your input will help shape the final implementation. Check the full list of improvements in the Python and Pylance changelogs.