PViz for VS Code
Thin VS Code wrapper around the pviz-parser CLI. Invokes pviz against the open workspace root, handles output path resolution, and surfaces the resulting bundle via native VS Code notifications. Configurable output path, analyzer mode, store root, and per-file size limits via VS Code settings. Full language coverage (Kotlin, Go, Rust) available via PViz cloud.
Requirements
Install the PViz CLI:
pip install pviz-parser
This installs the pviz command used by the extension.
Usage
- Open a workspace folder in VS Code
- Open the Command Palette (
Cmd+Shift+P / Ctrl+Shift+P)
- Run PViz: Analyze Workspace
- The bundle is saved to
.pviz/artifacts/bundle.json in your workspace by default
What the extension runs
pviz <workspace-root> -o <output-file> --mode zones --clean --allow-output-in-repo
--allow-output-in-repo is added automatically when the output file is inside the workspace. If you configure an external output path the flag is omitted. --clean ensures stale artifacts from previous runs on other repos are cleared.
Settings
| Setting |
Default |
Description |
pviz.outputFile |
.pviz/artifacts/bundle.json |
Output path for the bundle. Relative paths resolve from workspace root. |
pviz.cliPath |
pviz |
Path to the CLI if not on your PATH. |
pviz.mode |
zones |
Analyzer mode: zones or classic. |
pviz.maxBytes |
100000000 |
Maximum bytes per file. |
pviz.storeRoot |
(empty) |
Optional external store root. |
pviz.clean |
true |
Clears sandbox artifacts before each run to prevent stale cross-repo reuse. |
pviz.additionalArgs |
[] |
Extra arguments passed directly to the CLI. |
Supported Languages
Python, JavaScript, TypeScript, and Java (partial resolution) are supported by the local pviz-parser package. Full language coverage including Kotlin, Go, and Rust is available via PViz cloud.
What is a PViz bundle?
A structured JSON file containing your repo's dependency graph with centrality scoring and architectural analysis — optimized for use as LLM context. Drop it into Claude, GPT, or any long-context model to get architecture-aware responses about your codebase.