telnety.

MCP Server

Telnety includes a built-in Model Context Protocol (MCP) server that enables AI assistants like Claude to securely interact with your connections, hosts, and tools.

Overview

The MCP server communicates via JSON-RPC 2.0 and exposes Telnety operations as tools that AI assistants can call. Each operation is classified into one of four safety tiers:

Tier 1: Safe

Read-only operations. List hosts, read configurations, view sessions.

Tier 2: Moderate

Create connections, open sessions, transfer files.

Tier 3: Sensitive

Execute commands on remote hosts, modify system configurations.

Tier 4: Critical

Vault access, destructive operations. Requires explicit user approval.

Rate Limiting

  • 30 requests per minute (total)
  • 10 destructive operations per hour
  • 5 concurrent operations (semaphore)

Output Scrubbing

The MCP server automatically redacts sensitive data from command output before sending it to the AI assistant:

  • AWS access keys and secret keys
  • GitHub personal access tokens
  • PEM private keys
  • Passwords in URLs

Audit Logging

Every MCP operation is logged with session ID, host ID, action, safety tier, result, exit code, and execution time. Audit logs are stored in the local SQLite database and can be exported for compliance.