MCP Servers

A collection of Model Context Protocol servers, templates, tools and more.

M
Mcpgen Go
by @wl4g-ai
GitHub Repo(1 stars)

The Go MCP server Generator from OpenAPI 3.x specification.

Created 4/23/2026
Updated about 3 hours ago
README
View Source
Repository documentation and setup instructions

Go MCP server Generator from OpenAPI 3.x specification

Generate production-ready Model Context Protocol (MCP) servers from OpenAPI specs. Each API operation becomes an AI tool that forwards requests to your upstream service.

Building

make
# binary: bin/mcpgen

Quick Start

1. Generate the MCP server

./bin/mcpgen -i testdata/example_confluence_oas_v3.1.yaml -o myconfluence-mcp
cd myconfluence-mcp

This produces a complete Go project with tools for every operation:

myconfluence-mcp/
├── .credentials                 # file-based token (set MCP_UPSTREAM_TOKEN_FILE)
├── main.go                      # entry point (stdio/http transport)
├── client.sh                    # quick curl-based test script
├── Makefile                     # build / run / clean / test
├── myconfluence-mcp             # compiled binary
└── internal/
    ├── mcpserver/server.go      # MCP server setup + tool registration
    ├── helpers/                 # ForwardRequest, logging, parameter parsing
    └── mcptools/                # one file per API operation
        ├── GetPage.go
        ├── CreatePage.go
        ├── UpdatePage.go
        ├── DeletePage.go
        ├── SearchContent.go
        └── ...

2. Start the server

The server defaults to httpbin.org which echoes requests — great for quick verification:

./myconfluence-mcp --transport http --port 8080 -v 1
# MCP_UPSTREAM_ENDPOINT=https://httpbin.org/anything

Set your actual upstream to enable real API calls:

export MCP_UPSTREAM_ENDPOINT=https://example.atlassian.net/wiki/rest/api
# Option 1: pass token via env var
MCP_UPSTREAM_TOKEN=your-token ./myconfluence-mcp --transport http --port 8080 -v 1

# Option 2: read token from file (safer, no shell history exposure)
echo -n "your-token" > .credentials
MCP_UPSTREAM_TOKEN_FILE=.credentials ./myconfluence-mcp --transport http --port 8080 -v 1

3. Test with client.sh

./client.sh list-tools
./client.sh call GetPage '{"id": "123456"}'

Agent Integration

Local Mode (stdio)

Run the MCP server as a child process — recommended for local development.

OpenCode

~/.config/opencode/config.json:

{
  "mcp": {
    "myconfluence": {
      "type": "local",
      "command": ["bash", "-c", "./myconfluence-mcp"],
      "args": ["--transport", "stdio"],
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      },
      "enabled": true
    }
  }
}

Claude Code

~/.claude/settings.json:

{
  "mcpServers": {
    "myconfluence-r": {
      "command": "./myconfluence-mcp",
      "args": ["--transport", "stdio"],
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      }
    }
  }
}

Claude Desktop

~/.config/claude-desktop/claude_desktop_config.json:

{
  "mcpServers": {
    "myconfluence-r": {
      "command": ["bash", "-c", "./myconfluence-mcp"],
      "args": ["--transport", "stdio"],
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      }
    }
  }
}

Codex CLI

~/.codex/config.yaml:

mcp:
  servers:
    myconfluence-r:
      command: ./myconfluence-mcp
      args: ["--transport", "stdio"]
      env:
        MCP_UPSTREAM_ENDPOINT: https://example.atlassian.net/wiki/rest/api
        MCP_UPSTREAM_TOKEN: your-token

Cursor

~/.cursor/mcp.json:

{
  "mcpServers": {
    "myconfluence-r": {
      "command": "./myconfluence-mcp",
      "args": ["--transport", "stdio"],
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      }
    }
  }
}

Remote Mode (HTTP)

Run the server separately and connect agents via HTTP transport. Suitable for shared instances, cloud deployments, or when agent cannot spawn local processes.

Start the server:

export MCP_UPSTREAM_ENDPOINT=https://example.atlassian.net/wiki/rest/api
export MCP_UPSTREAM_TOKEN=your-token
./myconfluence-mcp --transport http --port 8080 -v 1

OpenCode (remote)

~/.config/opencode/config.json:

{
  "mcp": {
    "myconfluence-r": {
      "type": "remote",
      "url": "http://localhost:8080/mcp",
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      }
    }
  }
}

Claude Code (remote)

~/.claude/settings.json:

{
  "mcpServers": {
    "myconfluence-r": {
      "url": "http://localhost:8080/mcp",
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      }
    }
  }
}

Claude Desktop (remote)

~/.config/claude-desktop/claude_desktop_config.json:

{
  "mcpServers": {
    "myconfluence-r": {
      "url": "http://localhost:8080/mcp",
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      }
    }
  }
}

Codex CLI (remote)

~/.codex/config.yaml:

mcp:
  servers:
    myconfluence-r:
      url: http://localhost:8080/mcp
      env:
        MCP_UPSTREAM_ENDPOINT: https://example.atlassian.net/wiki/rest/api
        MCP_UPSTREAM_TOKEN: your-token

Cursor (remote)

~/.cursor/mcp.json:

{
  "mcpServers": {
    "myconfluence-r": {
      "url": "http://localhost:8080/mcp",
      "env": {
        "MCP_UPSTREAM_ENDPOINT": "https://example.atlassian.net/wiki/rest/api",
        "MCP_UPSTREAM_TOKEN": "your-token"
      }
    }
  }
}

Server CLI

| Flag | Description | Default | |---|---|---| | --transport <stdio\|http> | Transport mode | stdio | | --port <number> | HTTP server port | 8080 | | -v, --verbose <0-10> | Request logging verbosity | 0 |

Logging levels

| Level | Output | |---|---| | 0 | Silent | | 1 | HTTP access log: [http] sid=- 200 POST /mcp (1ms) | | 2 | MCP request log: [mcp] tool=SearchContent args={...}, upstream method + URL | | 3 | + upstream query params | | 5 | + request/response headers | | 7 | + request/response body | | 9 | + pretty-printed JSON body | | 10 | Same as 9 (full debug) |

Environment variables

| Variable | Description | |---|---| | MCP_UPSTREAM_ENDPOINT | Base URL of the upstream API (default: https://httpbin.org/anything) | | MCP_UPSTREAM_TOKEN | Bearer token for upstream auth (fallback when no Authorization header from client) | | MCP_UPSTREAM_TOKEN_FILE | Path to a file containing the bearer token (alternative to MCP_UPSTREAM_TOKEN) |

Token retrieval priority

The server tries to obtain a Bearer token in this order:

  1. Authorization header from the client's HTTP request (forwarded)
  2. MCP_UPSTREAM_TOKEN environment variable
  3. MCP_UPSTREAM_TOKEN_FILE (read from file — ideal for Kubernetes secrets)
  4. macOS Keychain (security find-generic-password -s mcpgen-upstream -wa "")
  5. Windows Credential Manager (cmdkey /get:mcpgen-upstream)

License

MIT License

Quick Setup
Installation guide for this server

Installation Command (package not published)

git clone https://github.com/wl4g-ai/mcpgen-go
Manual Installation: Please check the README for detailed setup instructions and any additional dependencies required.

Cursor configuration (mcp.json)

{ "mcpServers": { "wl4g-ai-mcpgen-go": { "command": "git", "args": [ "clone", "https://github.com/wl4g-ai/mcpgen-go" ] } } }