Vexp — Context Engine for AI Coding Agents

58% Less Cost. 90% Fewer Tool Calls. 100% On Your Machine.

SWE-bench Verified: 73% pass@1 at $0.67/task · Docs · Changelog

The Problem

AI coding agents waste more than half their budget before writing a single line of code. They read files, run searches, open dependencies — 23 operations on average just to orient themselves. The result: higher costs, slower responses, and context windows flooded with irrelevant code.

And when the session ends? Everything is forgotten. Tomorrow, the same agent rediscovers the same codebase from scratch.

What vexp Does

vexp pre-indexes your codebase into a dependency graph and delivers only the code that matters — ranked by relevance, bounded to your token budget. One install, and every AI agent you use starts informed instead of exploring blind.

• 23 tool calls → 2. The agent stops guessing and starts coding.
• $0.78 → $0.33 per task. Measured on SWE-bench Verified.
• 117s → 74s per task. Faster because there's less to process.

No accounts. No API keys. No code leaving your machine.

Benchmark Results

Tested on SWE-bench Verified — 100 real-world bugs, 4 agents, same model (Claude Opus), same budget.

On 7–10 tasks, vexp was the only agent to produce a passing patch. No other agent could solve them.

Full methodology and reproduction steps →

Quick Start

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

No account needed. No CLI. No API key. Works in under 60 seconds.

Works With 14+ AI Agents

vexp auto-detects and auto-configures every major AI coding agent via MCP. On activation, it generates agent-specific instructions — no manual setup.

Your agents don't compete with each other — they all share the same context layer.

How It Works

Your AI agent asks: "How does authentication work?"

Without vexp:
  → 23 tool calls, reads 14 files, 18,000 tokens consumed
  → Cost: $0.78 · Duration: 117s

With vexp:
  → 1 call to run_pipeline
  → Returns: 3 key functions (full code) + 8 related signatures (skeletons)
  → 2,400 tokens · Cost: $0.33 · Duration: 74s

Same question. 87% fewer tokens. Better answer.

Everything runs on your machine. No server. No cloud. No network calls.

Session Memory

Your AI doesn't start from scratch every time.

vexp observes what your agent explores, decides, and learns — and saves it across sessions. When code changes, outdated context is automatically flagged. No manual work. No black-box retrieval: every result explains why it was ranked.

• Auto-capture — every interaction is recorded as a compact observation
• Stale knowledge alerts — when code changes, related observations are flagged [STALE]
• Cross-session recall — pick up exactly where you left off, even days later
• Shared across agents — memory works between Cursor, Claude Code, Copilot, and every agent you use

Learns From Your Workflow

vexp builds intelligence from how you work — without you doing anything.

Every code change is tracked at the structural level: not just "file changed", but which functions were added, removed, or modified. vexp connects these changes to your agent's activity to surface what matters.

• Pattern Detection — recurring changes in the same area are surfaced as project-level guidelines
• Anti-Pattern Alerts — dead-end explorations and excessive edits are flagged before they waste your time
• Zero Configuration — all of this happens in the background

Smart Features

These work automatically — no configuration needed:

• Semantic Search — finds validateCredentials() when you search for "authentication", even if the word never appears in the function
• Intent Detection in 15 Languages — ask in English, Italian, French, German, Spanish, Portuguese, Russian, Chinese, Japanese, Korean, Arabic, Hindi, Turkish, Dutch, or Polish. "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
• Adaptive Results — repeated queries automatically expand the result budget
• Security Filter — .env, credentials, SSH keys are automatically excluded

Local AI — Optional +30% Savings

Turn on vexp-devmind — a code-specialised AI model that runs entirely on your machine — for an extra +30% token savings on top of the baseline.

• 100% local — the model never calls out. Your code stays on your hardware.
• GPU-accelerated — Metal on Apple Silicon, CUDA on NVIDIA. Auto-detected.
• One-click install — the sidebar prompts you once; ~3.5 GB one-time download.
• Always up to date — "Check LLM updates" compares your local version to the published release without downloading anything.

Sharper context ranking, higher-quality session memory, smarter prompt preprocessing. All without sending a single byte to any server.

30 Languages Supported

Covers ~99% of professional software development.

VS Code Integration

Beyond AI context, vexp adds visual intelligence directly in your editor:

• Sidebar Panel — daemon status, index stats, token savings, per-repo cards
• CodeLens — inline annotations on exported symbols: "N dependents across M files"
• Hover Info — impact data on exported declarations
• Status Bar — real-time indexing progress and daily usage counter

Multi-Repo Workspaces

Index multiple repositories as a unified workspace — entirely on your machine.

• Context queries span all repos automatically
• Cross-repo dependencies detected and navigable
• File changes tracked across repos with automatic correlation
• Ideal for monorepos, frontend+backend splits, and microservices

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
• .gitignore respected — only indexes files your project tracks

Your code never touches a server. Not ours. Not anyone's.

Platform Support

Native binaries — no runtime dependencies:

GPU acceleration applies to the optional local AI layer. Vexp's core context engine runs on any modern CPU.

Pricing

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

Pro pays for itself in 2 days. Average savings: $0.45/task × 20 tasks/day = $9/day vs $19/month.

CLI Available

vexp works from the terminal too. Use it with Claude Code, Codex, or any MCP-compatible agent — no VS Code required.

npm install -g vexp-cli
vexp setup

Same tools. Same languages. Same memory. Same intelligence.

→ vexp-cli on npm

Learn More

• Benchmark Results
• Documentation
• Changelog
• Website