secure-github-ops
Enables safe GitHub automation by queuing write operations for admin approval before execution.
README
Secure GitHub Ops MCP Server
Created by Aaron Murillo
An MCP server that makes GitHub automation actually safe. Every write operation needs approval before it runs, so you never have to worry about runaway scripts creating 100 issues while you're grabbing coffee.
What This Does
You know how most automation just... does stuff immediately? This flips that. When you (or an AI) try to create an issue, open a PR, or add a comment, it goes into a pending queue. Then someone with admin access reviews and approves it. Only then does it actually hit GitHub.
Two roles:
- Operators can request actions
- Admins can approve, deny, and see what's pending
Everything gets logged so you have a full audit trail.
Getting Started
You'll need Node 18+ and a GitHub token.
cd secure-github-ops
npm install
cp .env.example .env
Open .env and fill in your stuff:
GITHUB_TOKEN=ghp_your_token_here
ADMIN_API_KEYS=some-secret-admin-key
OPERATOR_API_KEYS=some-secret-operator-key
Then run it:
npm run build
npm start
For dev mode with hot reload:
npm run dev
Using with Claude Desktop
Add this to your Claude Desktop config:
{
"mcpServers": {
"secure-github-ops": {
"command": "node",
"args": ["/path/to/secure-github-ops/dist/index.js"],
"env": {
"GITHUB_TOKEN": "ghp_your_token",
"ADMIN_API_KEYS": "your-admin-key",
"OPERATOR_API_KEYS": "your-operator-key"
}
}
}
}
The Tools
For operators and admins:
- github.issues.create - request to create an issue
- github.pr.open - request to open a PR
- github.comment.add - request to add a comment
Admin only:
- admin.approve_action - approve and run a pending action
- admin.deny_action - reject a pending action
- admin.list_pending_actions - see what's waiting
How It Works
Say an operator wants to create an issue:
// Operator requests it
{
"tool": "github.issues.create",
"arguments": {
"_authToken": "operator-key",
"owner": "myorg",
"repo": "myrepo",
"title": "Fix the login bug"
}
}
// Returns a pending action with an ID
Nothing happened on GitHub yet. Now an admin checks what's pending:
{
"tool": "admin.list_pending_actions",
"arguments": { "_authToken": "admin-key" }
}
// Shows the pending issue request
Admin approves it:
{
"tool": "admin.approve_action",
"arguments": {
"_authToken": "admin-key",
"actionId": "the-uuid-from-before"
}
}
// NOW the issue gets created
Config Options
GITHUB_TOKEN (required) - Your GitHub PAT with repo access
ADMIN_API_KEYS (required) - Comma-separated admin keys
OPERATOR_API_KEYS (optional) - Comma-separated operator keys
ALLOWED_REPOS (optional) - Lock it down to specific repos like owner/repo1,owner/repo2
LOG_LEVEL (optional) - debug, info, warn, error
About the STDIO Transport
This uses STDIO, which means it talks through stdin/stdout. All the logs go to stderr so they don't mess with the protocol. If you're wondering why there's no console.log anywhere, that's why.
Want HTTP instead? That's on the roadmap. The core code stays the same, just swap the transport.
Running Tests
npm test
26 tests covering auth, config, and the pending store.
Docker
docker build -t secure-github-ops .
docker run -e GITHUB_TOKEN=... -e ADMIN_API_KEYS=... secure-github-ops
What's Next
Eventually adding HTTP transport so you can host this as a service. Also thinking about persistent storage for the pending actions (right now they live in memory).
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.