Skip to main content

MCP Vault: Secure Gateway & Booster for Antigravity

Project description

⚡ MCP Vault (mcpv)

The Ultimate Performance Booster for AI Agents
"Reduce system lag by 99%, eliminate loading times, and cut token costs by 90%."

License Python Platform Status


[!CAUTION] ⚠️ Compatibility Warning
Currently, this project ONLY supports Windows OS and the Antigravity agent environment.


❓ Why mcpv?

Have you ever felt this while using AI Agents (Antigravity, Cursor)?

"Why is it so heavy?"
"It froze again..."
"Why are the token costs so high?"

mcpv is not just a tool. It is a Turbo Engine for your agent.


🏎️ Overwhelming Performance Difference

Feature 😫 Without mcpv (Before) ⚡ With mcpv (After) 📈 Effect
Speed No GPU, Laggy UI Forced GPU Acceleration, Smooth 100x Perceived Speed
Loading Wait 60s+ every time 0.1s Instant Start (Lazy Load) Zero Latency
Cost Resend full code every time Auto-block Duplicates (Smart Cache) 90% Savings


✨ 3 Core Features

1️⃣ Booster Injection (Physical Acceleration)

"Unlock hardware limits with one line"

  • Forced GPU Activation: Injects hidden rendering acceleration flags (--enable-gpu-rasterization).
  • Permission Bypass: Drops Admin rights to fix drag-and-drop & UI bugs, and bypasses permission requests (Error 740) using RunAsInvoker.
  • Zombie Process Killer: Automatically cleans up ghost processes occupying ports.

2️⃣ Smart Valve (Cost Defense)

"Smart wallet protector that saves for you"

  • Detects the massive context data (repomix) that agents habitually request.
  • First request: Allowed (Full context provided).
  • Subsequent requests: Blocked with a 10-token "Already cached" message.
  • Physically blocks accidental token bombs.

3️⃣ Gateway Hijacking (Secure Vault)

"Stop struggling with complex configs"

  • Zero-Latency Startup: Only scans directories when the agent actually requests them. No timeouts on large repos.
  • Automatically migrates existing complex MCP settings to a secure Vault.
  • Original config is safely backed up to mcp_config.original.json.
  • The agent talks only to mcpv, but all tools work perfectly behind the scenes.


🛠️ Verified Recommended Setup

Verified MCP server configuration used by the developer. It creates the best synergy when used with mcpv.

{
  "mcpServers": {
    "rube": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://rube.app/mcp"]
    },
    "open-aware": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://open-aware.qodo.ai/mcp"]
    },
    "context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp", "--api-key", "PUT_IN_YOUR_API_KEY_HERE"]
    },
    "sequential-thinking": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-sequential-thinking"]
    },
    "mcp-server-neon": {
      "disabled": false,
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.neon.tech/sse"],
      "env": {
        "NEON_API_KEY": "NEVERCHANGE_DONT_PUT_IN_ANYTHING_ELSE_THAN_ME_HERE"
      }
    }
  }
}


� Recommended System Rule

Add this rule to your agent's system prompt (e.g., .cursorrules, .antigravityrules).
This ensures the agent uses mcpv correctly without redundant requests.

[System Instruction]

Session Start: At the very beginning of the conversation, you MUST call get_initial_context tool immediately. This is your only way to understand the project structure.

Context Management: Do NOT call get_initial_context again in the same session unless the user explicitly asks for a "refresh" with force=True. The proxy will block redundant requests.

File Access: After reading the initial context map, use read_file to inspect specific files needed for the task.



�🛠️ Tech Stack

  • Python & uv: Uses uv for clean, isolated virtual environments to prevent dependency hell.
  • FastMCP: Built on top of the FastMCP framework for speed.
  • Gateway Pattern: It backs up your original config and routes traffic through the proxy securely.


📦 Installation

Choose the method that fits your needs.

Option A: The "It Just Works" Method (Recommended for most users)

If you have Python installed and added to PATH.

# 1. Install from PyPI
pip install mcpv

# 2. Setup Gateway & Booster
mcpv install

# 3. Done! 
# Look for the "Antigravity Boost (mcpv)" shortcut on your Desktop.

Option B: The "Rock-Solid" Method (Using uv)

Use this method if your Python environment is messy or you want complete isolation.

1. Clean up existing processes

Stop-Process -Name "mcpv" -Force -ErrorAction SilentlyContinue
Stop-Process -Name "python" -Force -ErrorAction SilentlyContinue

2. Create Virtual Environment & Install

# Install uv (if needed)
powershell -ExecutionPolicy ByPass -c "irm [https://astral.sh/uv/install.ps1](https://astral.sh/uv/install.ps1) | iex"

# Create .venv and install mcpv
uv venv
uv pip install mcpv

3. Register & Lock

This registers the isolated virtual environment into Antigravity's config.

.venv\Scripts\python -m mcpv install --force

4. Run

Launch Antigravity Boost (mcpv) from your Desktop.


🛠️ Commands

Command Description
mcpv install Installs the gateway and creates the Desktop shortcut.
mcpv install --force Forces installation even if only 1 MCP server exists.
mcpv start Starts the server (Used internally by Antigravity).
mcpv --help Show help message.

❓ Troubleshooting

Q. "File access denied" or "Failed to remove file" during update?

This means mcpv.exe or Antigravity is still running.

  1. Close Antigravity.
  2. Run: Stop-Process -Name "mcpv" -Force in PowerShell.
  3. Try installing again.

Q. I installed it, but I don't see the Shortcut.

The shortcut is created on your Desktop. If not, check the installation logs. You can create it manually by running mcpv install again.


📜 License

MIT License. Feel free to fork and modify!


Support
If this project helped you save tokens and time, consider buying me a coffee!

Buy Me A Coffee



⚡ Charged by MCP Vault
Developed for High-Performance AI Agent Operations

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

mcpv-0.2.6.tar.gz (28.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

mcpv-0.2.6-py3-none-any.whl (11.5 kB view details)

Uploaded Python 3

File details

Details for the file mcpv-0.2.6.tar.gz.

File metadata

  • Download URL: mcpv-0.2.6.tar.gz
  • Upload date:
  • Size: 28.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for mcpv-0.2.6.tar.gz
Algorithm Hash digest
SHA256 9029f63f237e5b9a66622a8cff9a15a78dcbbd44b6b036ab8b019d64d53a15de
MD5 b6cbef3784ea9d77a272d98fcddeb889
BLAKE2b-256 9c91b876e65434088abdb380ea403b4988464ca3a15cd27e52b58b744640cad4

See more details on using hashes here.

File details

Details for the file mcpv-0.2.6-py3-none-any.whl.

File metadata

  • Download URL: mcpv-0.2.6-py3-none-any.whl
  • Upload date:
  • Size: 11.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for mcpv-0.2.6-py3-none-any.whl
Algorithm Hash digest
SHA256 d59ba4e44e0ab7f3d28e2cc1a6e83e153c0fdfe46a30706156513b23ef2477c0
MD5 6f469a8fd9b35965bf73e4845f8779cd
BLAKE2b-256 cd8d9c66fb0fa44968c75571e2af10fe3e9721743f1739df13c0952ce8f1c215

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page