vexp

The Local-First Context Engine for AI Coding Agents

Save 70% of tokens. Get 2x better answers. Your code never leaves your machine.

The Problem

AI coding agents read entire files, dump them into context, and burn through tokens fast. On a 50-file project, a single question can consume 30k+ tokens of irrelevant code. Responses suffer. Costs go up.

Cloud-based context engines solve this — but they require your code to leave the building.

The Solution

vexp builds a live dependency graph of your codebase using advanced static analysis — entirely on your machine. When an AI agent asks a question, vexp returns only the pivot nodes that matter — the exact functions, classes, and types relevant to the task — plus compact skeletons of surrounding context.

The result: 65-70% fewer tokens, faster responses, and dramatically more accurate answers. No accounts. No API keys. No code leaving your laptop.

Quick Start

1. Install the vexp extension from the VS Code Marketplace
2. Open any project folder — vexp indexes your codebase automatically
3. Ask your AI agent anything — vexp delivers the right context, automatically

No account needed. No CLI to install. No API key to configure.

Works With 12+ AI Agents

vexp auto-detects and auto-configures every major AI coding agent via MCP (Model Context Protocol). On activation, vexp generates the agent-specific config file with tool usage instructions tailored to each agent.

Install the extension, open your project, and your AI agent instantly gains access to vexp's context engine. No manual setup required.

11 Languages Supported

Covers ~90% of professional software development. Each language has a dedicated grammar with language-specific extraction rules.

How It Works

1. Open your project — vexp analyzes your codebase in seconds
2. Your AI agent asks: "How does authentication work?"
3. vexp returns a Context Capsule:
   - 3 pivot functions (full code)
   - 8 supporting skeletons (signatures only)
   - Total: 2,400 tokens instead of 18,000

Same question. 87% fewer tokens. Better answer.

What's Inside — 10 MCP Tools

Session Memory

Your AI doesn't start from scratch every time.

vexp remembers what your agent explored, decided, and learned — across sessions, across repos. When code changes, outdated memories are automatically flagged. No black-box retrieval: every result explains why it was ranked.

The only context engine with persistent, code-aware memory.

Smart Features

These features work automatically — no configuration needed:

• Smart Search — finds validateCredentials() when you search for "authentication", even if the word never appears in the function
• Intent Detection — "fix bug" focuses on error paths, "refactor" returns blast-radius analysis, "add feature" prioritizes module boundaries
• Change Awareness — files frequently modified together are automatically linked in the graph
• Code Hotspots — actively maintained code gets a ranking boost
• Adaptive Results — repeated queries automatically expand the result budget
• Security Filter — sensitive files (.env, credentials, SSH keys) are automatically excluded from the index

VS Code Integration

Beyond the MCP tools that AI agents use, vexp provides visual features directly in VS Code:

• Sidebar Panel — daemon status, index statistics, per-repo cards, action buttons
• CodeLens — inline annotations on exported symbols: "N dependents across M files"
• Hover Info — impact data on exported declarations (dependents, cross-repo count, top callers)
• Status Bar — real-time indexing progress and daemon health

Multi-Repo Workspaces

Index multiple repositories as a unified workspace — entirely on your machine. Ideal for monorepos, frontend+backend splits, and microservice architectures.

• Context queries span all repos automatically
• Cross-repo dependencies are detected and navigable
• Open any repo in the workspace and all repos are available

Security

• 100% local processing — zero network calls, zero cloud dependencies
• Binary verification — SHA-256 checksums verified on extension activation
• Secret filtering — sensitive files and content patterns automatically excluded
• Custom exclusions — .vexp_ignore for additional file exclusions

Platform Support

Native binaries — no runtime dependencies:

Pricing

All plans run 100% locally. No per-query costs, no credit metering, no surprise bills.

Learn More

Visit vexp.dev for documentation, guides, and support.