Semantic search for your VS Code workspace using local AI embeddings. Find code by concept, not just keywords — all without sending your code to a server.
- Semantic vector search — find code related to a concept or behaviour, even when the exact words don't match
- Fully local — embeddings are computed in-browser using Transformers.js; no data leaves your machine
- Copilot tool integration — expose search to GitHub Copilot Chat, and any other chat, so it can query your codebase with
#intellisearch - Incremental index updates — file watchers keep the index up to date as you edit
- Open the Command Palette (
Ctrl+Shift+P/Cmd+Shift+P) - Run IntelliSearch: Build Index
The panel opens and indexes every file in your workspace. This takes a moment on first run while the AI model loads and files are embedded.
- Use the search box in the IntelliSearch panel, or
- Reference
#intellisearchin a GitHub Copilot chat message to let the AI search on your behalf
Run IntelliSearch: Build Index again at any time to do a full rebuild. The index is updated automatically in the background as you save files.
No settings yet.
- Large workspaces may take several minutes to index on the first build