dev-kit

Dev Kit is an MCP server that bridges AI models with a suite of development tools, enabling natural language interactions to manage Jira issues, Confluence pages, and GitLab/GitHub repositories and workflows. It supports two communication protocols: STDIO (default) and SSE for real-time streaming. After starting the server, you can instruct the model to perform tasks such as searching and creating Jira issues, reading and updating Confluence pages, listing and manipulating GitHub/GitLab repositories, issues, and merge requests, and more, all through the defined tool groups. Use the ENABLE_TOOLS environment variable to limit which tool groups are available, or leave it empty to enable all tools by default. The server reads configuration from a .env file you provide, containing credentials and host information for Atlassian, GitLab, and GitHub services, as well as any network proxy or port settings.

Prerequisites:

• Go 1.23.2 or higher
• Access tokens and credentials for Atlassian (Jira & Confluence), GitLab, and GitHub as needed
• A host environment where you can run Go binaries

Installation steps:

1. Install the server using Go:

go install github.com/nguyenvanduocit/dev-kit@latest

2. Prepare your environment configuration. Create a .env file containing credentials and hosts, for example:

# Atlassian
ATLASSIAN_HOST=https://your-domain.atlassian.net
ATLASSIAN_EMAIL=you@example.com
ATLASSIAN_TOKEN=your-atlassian-api-token

# GitLab
GITLAB_HOST=https://gitlab.example.com
GITLAB_TOKEN=your-gitlab-token

# GitHub
GITHUB_TOKEN=your-github-token

# Optional
ENABLE_TOOLS=confluence,github,gitlab,jira,script
PROXY_URL=http://proxy.example.com:8080
PORT=8080

3. Run the server via MCP config. The README shows a sample mcpServers entry like:

{
  "mcpServers": {
    "dev_kit": {
      "command": "dev-kit",
      "args": [
        "-env", "/path/to/.env",
        "-protocol", "stdio"
      ]
    }
  }
}

If you want SSE, set -protocol sse and ensure PORT is set in the environment.

4. Start the server and connect your MCP consumer (e.g., an AI agent) to it using the protocol you chose.

Tips and notes:

• Ensure the .env path is accessible by the server process and that all required keys are set.
• If you encounter authentication errors, verify token scopes for Jira, Confluence, GitLab, and GitHub.
• The ENABLE_TOOLS setting can help reduce surface area by restricting which tool groups are exposed to the model.
• For SSE, make sure the specified PORT is open and not blocked by a firewall; defaults to 8080 if PORT is not set.
• The server is designed to automate development workflows and may perform actions on your projects; use with appropriate permissions and test in a safe environment.
• If you modify tool groups or environment variables, restart the MCP server to apply changes.