mcp-keyward
Local encrypted vault for API keys. LLMs never see your real keys ā the server injects them transparently into HTTP requests.
README
š mcp-keyguard
Your AI agent should never see your API keys. Now it won't.
mcp-keyguard is a local MCP server that acts as a secure proxy between your AI agent (Claude, Cursor, Windsurf...) and any external API.
Instead of pasting your OpenAI, Stripe or GitHub keys into the chat context, you store them encrypted on your machine. The agent calls mcp-keyguard, which injects the real key server-side and returns the result.
The key never leaves your machine. The agent never sees it.
Why this matters
In 2025, a vulnerability in a popular MCP hosting platform exposed thousands of API keys from over 3,000 servers. The root cause? Keys passed through infrastructure the user didn't control.
mcp-keyguard is the opposite: fully local, zero external dependencies, your keys encrypted at rest with AES-128.
How it works
Your prompt ā Claude ā mcp-keyguard ā [injects real key] ā External API
ā
Key never leaves here
Installation
Requirements: Python 3.10+
pip install mcp httpx cryptography
git clone https://github.com/ggc180820/mcp-keyguard.git
cd mcp-keyguard
Add to your claude_desktop_config.json:
{
"mcpServers": {
"mcp-keyguard": {
"command": "python",
"args": ["/path/to/mcp-keyguard/main.py"]
}
}
}
Restart Claude Desktop. Done.
Usage
1. Store a key (you do this once per key)
"Use add_key to store my OpenAI key with alias 'openai',
header_name 'Authorization', header_prefix 'Bearer '"
2. Make authenticated requests (Claude does this automatically)
"Use make_request with alias 'openai' to call
https://api.openai.com/v1/models"
3. Check what's stored
"Use list_keys"
ā Shows aliases and headers. Never the real values.
Tools
| Tool | What it does |
|---|---|
add_key |
Store an API key encrypted in the vault |
list_keys |
List stored aliases ā values are never shown |
make_request |
Make an authenticated HTTP request, key injected server-side |
delete_key |
Remove a key from the vault |
Security model
- Keys are encrypted with Fernet (AES-128-CBC + HMAC-SHA256)
- The encryption key lives in
vault.keyon your machine only - All HTTP requests are made locally with a 30s timeout
- Never commit
vault.keyorvault.jsonto git (already in.gitignore)
mcp-keyguard Pro
Need more control? Pro adds:
| Feature | Free | Pro |
|---|---|---|
| Encrypted local vault | ā | ā |
| Unlimited keys | ā | ā |
| Multiple vaults (per project/client) | ā | ā |
| Audit log (who used which key, when, where) | ā | ā |
| Key rotation alerts | ā | ā |
License
MIT ā free forever for personal use.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.