strands-mcp-server 0.1.2

Transform Strands Agents into MCP servers, exposing agent tools to any MCP-compatible client

strands-mcp-server

Bidirectional MCP integration for Strands Agents. Expose agents as MCP servers or connect to any MCP server.

pip install strands-mcp-server

---

Overview

Two Python tools + CLI for Model Context Protocol (MCP):

• mcp_server: Expose your agent's tools as MCP server
• mcp_client: Connect to and use remote MCP servers
• CLI: stdio server for Claude Desktop/Kiro

graph TD
    A[Your Strands Agent<br/>tools: calculator, shell, file_read] --> B[mcp_server tool]
    B --> C[MCP Protocol<br/>HTTP/stdio]
    C --> D[MCP Clients]
    D --> E[Claude Desktop]
    D --> F[Other Agents]
    D --> G[Custom Clients]
    
    style A fill:#2d3748,stroke:#4a5568,color:#fff
    style B fill:#2b6cb0,stroke:#2c5282,color:#fff
    style C fill:#38a169,stroke:#2f855a,color:#fff
    style D fill:#805ad5,stroke:#6b46c1,color:#fff
    style E fill:#d69e2e,stroke:#b7791f,color:#fff
    style F fill:#d69e2e,stroke:#b7791f,color:#fff
    style G fill:#d69e2e,stroke:#b7791f,color:#fff

---

Quick Start

1. Expose Your Agent (Server)

from strands import Agent
from strands_tools import calculator, shell
from strands_mcp_server import mcp_server

agent = Agent(tools=[calculator, shell, mcp_server])
agent("start mcp server on port 8000")
# Server running at http://localhost:8000/mcp

Stateless mode (production):

agent.tool.mcp_server(action="start", stateless=True, agent=agent)

2. Connect to Servers (Client)

from strands import Agent
from strands_mcp_server import mcp_client

agent = Agent(tools=[mcp_client])

# Connect
agent.tool.mcp_client(
    action="connect",
    connection_id="server1",
    transport="http",
    server_url="http://localhost:8000/mcp"
)

# Call remote tool
agent.tool.mcp_client(
    action="call_tool",
    connection_id="server1",
    tool_name="calculator",
    tool_args={"expression": "42 * 89"}
)

3. Claude Desktop Integration

Local mode (expose ./tools/ directory):

{
  "mcpServers": {
    "my-tools": {
      "command": "uvx",
      "args": [
        "strands-mcp-server",
        "--cwd", "/absolute/path/to/project"
      ]
    }
  }
}

Proxy mode (bridge to HTTP server):

{
  "mcpServers": {
    "my-agent": {
      "command": "uvx",
      "args": [
        "strands-mcp-server",
        "--upstream-url", "http://localhost:8000/mcp"
      ],
      "autoApprove": ["tool_name"]
    }
  }
}

---

API Reference

mcp_server

Parameter	Type	Default	Description
action	str	required	start, stop, status, list
transport	str	http	http (background) or stdio (foreground)
port	int	8000	HTTP port
tools	list[str]	None	Tools to expose (None = all)
expose_agent	bool	True	Enable invoke_agent tool
stateless	bool	False	Multi-node ready (no session state)

Examples:

# Basic
agent("start mcp server")

# Specific tools only
agent.tool.mcp_server(action="start", tools=["calculator"], agent=agent)

# Production
agent.tool.mcp_server(action="start", stateless=True, agent=agent)

mcp_client

Parameter	Type	Description
action	str	connect, disconnect, list_tools, call_tool, list_connections
connection_id	str	Connection identifier
transport	str	http, stdio, sse
server_url	str	URL (http/sse)
tool_name	str	Tool to call
tool_args	dict	Tool arguments

Transports:

# HTTP
agent.tool.mcp_client(
    action="connect", connection_id="s1",
    transport="http", server_url="http://localhost:8000/mcp"
)

# stdio (subprocess)
agent.tool.mcp_client(
    action="connect", connection_id="s2",
    transport="stdio", command="python", args=["server.py"]
)

# SSE
agent.tool.mcp_client(
    action="connect", connection_id="s3",
    transport="sse", server_url="http://localhost:8000/sse"
)

invoke_agent Tool

When expose_agent=True (default), clients get full conversational access to the agent:

# Basic usage via mcp_client
agent.tool.mcp_client(
    action="call_tool",
    connection_id="remote",
    tool_name="invoke_agent",
    tool_args={"prompt": "Calculate 2 + 2 and explain your reasoning"}
)

Parameters:

• prompt (required): Natural language query for the agent

The agent will process the prompt using its configured model and tools, returning a complete response.

---

CLI

uvx strands-mcp-server [OPTIONS]

Option	Description
--cwd PATH	Working directory (local mode)
--upstream-url URL	Upstream server (proxy mode)
--system-prompt TEXT	Custom system prompt
--no-agent-invocation	Disable invoke_agent
--debug	Debug logging

Local mode: Exposes ./tools/ directory with hot reload Proxy mode: Bridges stdio ↔ HTTP server

Examples:

# Local mode
uvx strands-mcp-server --cwd /path/to/project

# Proxy mode
uvx strands-mcp-server --upstream-url http://localhost:8000/mcp

# With debug logging
uvx strands-mcp-server --cwd /path/to/project --debug

Troubleshooting

Tools not loading:

uvx strands-mcp-server --cwd /absolute/path --debug

Connection refused:

curl http://localhost:8000/mcp

Port in use:

lsof -i :8000
kill -9 <PID>

Check Claude Desktop logs:

tail -f ~/Library/Logs/Claude/mcp*.log

---

Testing

pytest              # All tests
pytest --cov        # With coverage

---

Links

• Docs: https://cagataycali.github.io/strands-mcp-server/
• Source: https://github.com/cagataycali/strands-mcp-server
• PyPI: https://pypi.org/project/strands-mcp-server/
• Strands: https://strandsagents.com
• MCP Spec: https://modelcontextprotocol.io/

---

License

Apache 2.0