mcp-shield
The nginx of MCP — drop-in resilience middleware for any MCP server.
mcp-shield
The nginx of MCP — drop-in resilience middleware for any MCP server.
Timeout · Retry · Circuit Breaker · Structured Logging
Why
MCP servers have zero built-in resilience. A hung GitHub API blocks your agent for 600 seconds. A transient network blip crashes the entire chain. A dead server keeps getting hammered with requests.
mcp-shield is a transparent stdio proxy that sits between your agent and MCP server. One command, zero code changes.
Agent ←stdio→ mcp-shield ←stdio→ MCP Server
Install
npm install -g @daino/mcp-shield
Or run directly with npx:
npx @daino/mcp-shield wrap -- npx @modelcontextprotocol/server-github
Quick Start
# Wrap any MCP server with sensible defaults (30s timeout, 2 retries)
mcp-shield wrap -- npx @modelcontextprotocol/server-github
# Custom timeout and retries
mcp-shield wrap --timeout 60s --retries 5 -- npx server-github
# Using a config file
mcp-shield wrap --config mcp-shield.yaml --server github
Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"github": {
"command": "npx",
"args": [
"@daino/mcp-shield", "wrap",
"--timeout", "30s", "--retries", "3",
"--",
"npx", "@modelcontextprotocol/server-github"
],
"env": {
"GITHUB_TOKEN": "your-token-here"
}
}
}
}
Done. Your GitHub MCP server now has timeout, retry, and circuit breaker protection.
Features
Timeout
Kill hung tool calls. No more 600-second waits.
timeout: 30s # per-tool override available
Retry
Exponential backoff + jitter. Deterministic errors (invalid params, method not found) are never retried.
retries:
max: 3
backoff: exponential # 1s → 2s → 4s
jitter: true
Circuit Breaker
After repeated failures, fail fast instead of burning tokens on a dead server.
circuit_breaker:
threshold: 5 # open after 5 consecutive failures
reset_after: 60s # try again after 60 seconds
States: Closed (normal) → Open (rejecting) → Half-Open (testing)
Structured Logging
Every tool call logged as structured JSON to stderr:
{
"level": "info",
"msg": "tool_call_end",
"server": "github",
"tool": "get_file_contents",
"duration_ms": 245,
"status": "success",
"attempt": 1
}
Config File
For multi-server setups:
# mcp-shield.yaml
defaults:
timeout: 30s
retries:
max: 3
backoff: exponential
jitter: true
circuit_breaker:
threshold: 5
reset_after: 60s
servers:
github:
command: "npx @modelcontextprotocol/server-github"
env:
GITHUB_TOKEN: "${GITHUB_TOKEN}"
tools:
get_file_contents:
timeout: 60s # slow tool gets more time
search_repositories:
retries:
max: 5 # flaky tool gets more retries
filesystem:
command: "npx @modelcontextprotocol/server-filesystem /home/user"
timeout: 10s
retries:
max: 1
Programmatic Usage
import { shield } from '@daino/mcp-shield';
const proxy = shield({
command: 'npx',
args: ['@modelcontextprotocol/server-github'],
timeout: 30_000,
retries: { max: 3, backoff: 'exponential', jitter: true },
circuitBreaker: { threshold: 5, resetAfter: 60_000 },
});
proxy.start();
Comparison
| mcp-shield | No protection | General retry libs | |
|---|---|---|---|
| MCP-native (JSON-RPC aware) | ✅ | — | ❌ |
| Per-tool config | ✅ | — | ❌ |
| Zero agent code changes | ✅ | — | ❌ |
| Circuit breaker | ✅ | ❌ | ✅ |
| Structured MCP logging | ✅ | ❌ | ❌ |
| Drop-in Claude Desktop support | ✅ | — | ❌ |
Roadmap
- Timeout + Retry + Circuit Breaker + Logging
- Response Validation (schema check)
- Tool Filtering (expose only specific tools)
- Rate Limiting (per-tool call caps)
- Metrics Export (Prometheus-compatible)
- Multi-server Composition
- Hot-reload config
- Dashboard UI
Contributing
Contributions welcome! Please open an issue first to discuss what you'd like to change.
git clone https://github.com/DainoJung/mcp-shield.git
cd mcp-shield
npm install
npm test
License
Related Servers
Google Sheets
Interact with Google Sheets using a Python-based MCP server and Google Apps Script.
Scenext MCP Server
Integrates with the Scenext AI platform to generate educational videos on various topics.
Promptheus
AI-powered prompt refinement tool with adaptive questioning and multi-provider support. Intelligently refines prompts through clarifying questions, supports 6+ AI providers (Google Gemini, Anthropic Claude, OpenAI, Groq, Alibaba Qwen, Zhipu GLM), and provides comprehensive prompt engineering capabilities.
Sequential Story
An MCP server for problem-solving using Sequential Thinking and Sequential Story mnemonic techniques.
Office PowerPoint MCP
Create, edit, and manipulate PowerPoint presentations using python-pptx.
UNO: Unified Narrative Operator
A text enhancement tool that transforms story content into rich, detailed narratives using advanced literary techniques and heuristic analysis.
ffl-mcp
Give AI a "send file" capability via P2P (Local-first)
LimeSurvey
Manage surveys and responses in your LimeSurvey instance.
Word MCP Server
Create and edit Microsoft Word (.docx) documents via an API.
HiveFlow
Connect AI assistants directly to the HiveFlow automation platform.