Local-first privacy
Project scanning, embeddings, the SQLite index, and semantic search run on your machine with local Ollama. No hosted embedding API is required for the indexing pipeline.
Local-first developer memory
Your codebase.
Your codebase.
Remembered.
Stop digging through folders and repeating context. ContextVault indexes your project locally and gives your AI tools a searchable memory of the code that matters.
Built for Windows 10 (build 19041+) and Windows 11 · Powered by local Ollama · No hosted embedding API required
ContextVault
Local workspace
Search your project context.
Find relevant code and knowledge across your indexed project.
Files indexed248
Code chunks1,426
RuntimeLocal
⌕Where is user authentication validated?
src / auth / middleware.tsMATCH 96.8%
export async function validateSession(request) {
const session = await readSession(request);
return verifyUser(session.userId);
src / api / routes.tsMATCH 91.2%
router.post("/login", validateCredentials);
Local-first by design
Works with Cursor & Claude Desktop
Semantic code search
Your files stay on your machine
Why ContextVault?
Project context, without sending your index to the cloud.
Keep retrieval close to your code, connect the tools you already use, and search indexed project context locally.
Works with Claude & Cursor
Connect through MCP and let compatible clients retrieve relevant project context with ContextVault's search and indexing tools.
Fast local retrieval
Search your local semantic index without a hosted embedding-service round trip for each query. Results are ranked from chunks already indexed on your machine.
A better way to keep context
Less searching. More building.
ContextVault turns your local project into searchable context, ready when you or your AI coding tools need it.
Local-first
Indexing, embeddings, and search run on your machine through local Ollama. Your project does not need a hosted embedding service.
Works with AI tools
Connect through the Model Context Protocol and expose project search and indexing tools to compatible clients such as Cursor and Claude Desktop.
Fast to find
Semantic search ranks the most relevant code chunks, so you can move from a question to the right part of your codebase faster.
Private by design
Project files and embeddings stay in your local ContextVault database and Ollama process. No automatic source-code upload to hosted embedding APIs.
Simple pricing
Start free. Upgrade when you need more.
Begin with ContextVault Free, then choose monthly or annual Pro billing through one checkout.
Free
$0 / month
Explore the local-first ContextVault workflow and start building searchable project context.
Free planPro
$9 / month
Or $79 / year with annual billing.
Upgrade to Pro ↗The Pro checkout lets you choose monthly or yearly billing. Subscription terms are shown before payment.
Get started in minutes
Your project memory, in three steps.
Install once, index a project, then search your code from the web interface or an MCP-compatible desktop client.
Install ContextVault
Install Ollama for Windows, then run the ContextVault installer. It prepares the supported Node.js runtime and project dependencies.
Index your project
Enter a local project folder in the onboarding panel. ContextVault scans supported files and creates embeddings locally.
Search and stay current
Ask natural-language questions about your code. Live monitoring adds new and changed files to your project index.
Questions, answered
Frequently asked questions
Does ContextVault upload my source code?
ContextVault performs project scanning, chunking, embedding generation, and search locally with Ollama. It does not automatically send project source files to a hosted embedding API. When you use an AI desktop client, that client receives the tool results you request.
Which tools can connect to ContextVault?
ContextVault exposes an MCP server with project indexing and semantic search tools. It can be configured with compatible MCP clients, including Claude Desktop and Cursor.
What do I need before using it?
Install Ollama for Windows before running the installer. ContextVault prepares Node.js 22 and project dependencies; the first launch downloads the nomic-embed-text embedding model if it is not already available.
Can ContextVault keep its index up to date?
Yes. After you index a project, the live watcher monitors file additions and changes and updates the local index.
Give your codebase a memory.
Start with ContextVault and keep project context close to the code.