MCP Catalogs
HomeSentinelgate screenshot

Sentinelgate

by Sentinel-Gate·25·Score 44

SentinelGate provides MCP proxy with policy enforcement and auditing for AI agent actions.

securitydeveloper-toolsops-infra
1
Forks
0
Open issues
1 mo ago
Last commit
2d ago
Indexed

Overview

SentinelGate is an access control system for AI agents that intercepts every action before execution. It works as an MCP proxy with a policy decision point, using CEL policies and RBAC to allow or deny tool calls, shell commands, and file operations. The system provides deterministic security controls and maintains a full audit trail of all agent activities.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:Preventing prompt injection attacks through external content
you:Blocking destructive commands during cleanup operations
you:Stopping unauthorized data exfiltration via API calls
you:What makes SentinelGate different from a regular firewall?
you:Does it require modifying my AI agent?

When to choose this

Choose SentinelGate when you need to control and audit AI agent access to sensitive systems, especially when agents operate in containers or sandboxes with unrestricted tool permissions.

When NOT to choose this

Don't choose SentinelGate if you're looking for an MCP server to expose tools/resources to agents, as it's actually a security proxy that controls agent actions, not a provider of MCP capabilities.

Tools this server exposes

12 tools extracted from the README
  • read_filepath: string

    Read a file from the filesystem

  • write_filepath: string, content: string

    Write content to a file

  • list_filespath: string

    List files and directories in a path

  • execute_bashcommand: string

    Execute a bash command

  • make_http_requestmethod: string, url: string, headers: object, body: string

    Make an HTTP request to a web service

  • delete_filepath: string

    Delete a file from the filesystem

  • create_directorypath: string

    Create a new directory

  • rename_filesource: string, destination: string

    Rename or move a file

  • get_file_infopath: string

    Get metadata information about a file

  • search_filespath: string, pattern: string

    Search for files matching a pattern

  • compress_filesfiles: array, output: string

    Compress multiple files into an archive

  • decompress_filearchive: string, destination: string

    Decompress an archive to a destination directory

Note: Tool names inferred from the MCP proxy behavior described in the README, as the README doesn't explicitly list all available tools but mentions '12 discovered' tools in the example startup output. The tools are common file system and system

Comparable tools

firewallopsgenieopen Policy AgentKubeflowAI guardrails

Installation

Installation

**macOS / Linux:**

curl -sSfL https://raw.githubusercontent.com/Sentinel-Gate/Sentinelgate/main/install.sh | sh

**Windows PowerShell:**

irm https://raw.githubusercontent.com/Sentinel-Gate/Sentinelgate/main/install.ps1 | iex

Claude Desktop Configuration

To use with Claude Desktop, add to your config.json:

{
  "mcpServers": {
    "sentinelgate": {
      "command": "sentinelgate",
      "args": ["serve"]
    }
  }
}

FAQ

What makes SentinelGate different from a regular firewall?
SentinelGate operates at the application layer specifically for AI agent actions, understanding context like tool calls and content, rather than just network traffic.
Does it require modifying my AI agent?
No. SentinelGate works as a proxy between the agent and your system without requiring any code changes to the agent itself.

On Hacker News

Recent discussion from the developer community.

Compare Sentinelgate with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.