native-mcp

Connect to MCP servers and register their tools as callable functions.

Updated May 15, 2026
One-click install
npx skills add https://github.com/cabezno/bmb-encover-agent --skill native-mcp-cabezno
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: native-mcp
Source: https://github.com/cabezno/bmb-encover-agent/tree/main/skills/mcp/native-mcp
Command: npx skills add https://github.com/cabezno/bmb-encover-agent --skill native-mcp-cabezno

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires mcp.

What problem does it solve?

Native MCP client solves the problem of integrating external tool providers into the Encover agent without writing custom glue code or maintaining separate bridge workflows.

Core Features & Use Cases

  • Connect to MCP servers at startup: Discovers available tools from configured MCP servers and makes them available as first-class agent tools.
  • Support for stdio and HTTP transports: Run local MCP servers via command (stdio) or connect to remote MCP servers over HTTP/StreamableHTTP.
  • Predictable tool naming and safe execution: Registers tools using the mcp_{server_name}_{tool_name} convention and applies environment filtering and credential redaction to reduce sensitive data leakage.
  • Built for persistent, always-on tool access: Keeps MCP connections alive across conversations and reconnects on failures with backoff.

Quick Start

Add your MCP servers to ~/.bmb/config.yaml under the mcp_servers key, restart the agent, then ask it to use the discovered MCP tool(s).

Frequently Asked Questions about native-mcp

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I connect an agent to MCP servers for tool discovery?

The native-mcp Skill connects the Encover agent to configured MCP servers, discovers their tools, and registers them as callable first-class tools without writing custom glue code or maintaining separate bridge workflows.

Can I use both stdio and HTTP transports for MCP servers?

Yes, MCP servers support both stdio and HTTP transports. You can run local servers via command using stdio or connect to remote MCP servers over HTTP and StreamableHTTP protocols.

How are discovered MCP tools named and registered?

Discovered MCP tools are registered using the mcp_{server_name}_{tool_name} naming convention. The system also applies environment filtering and credential redaction to reduce sensitive data leakage during execution.

Do MCP connections stay alive across conversations or reconnect on failure?

MCP connections are built for persistent, always-on tool access. They stay alive across conversations and automatically reconnect on failures using a backoff strategy to maintain tool availability.

Does the MCP client support environment filtering and credential redaction?

Yes, the MCP client applies environment filtering for stdio subprocesses and credential redaction to reduce sensitive data leakage. These features are applied automatically during tool discovery and execution.