MCP Architecture<\/strong><\/h3>\n\n\n\nMCP<\/a> uses a client-server model where your AI application (the MCP host<\/strong>) coordinates and manages one or multiple MCP clients<\/strong> that connect one-to-one with MCP servers<\/strong>, like dedicated phone lines for each tool.<\/p>\n\n\n\nTwo layers power this:<\/p>\n\n\n\n
1. Data layer<\/strong><\/h4>\n\n\n\nThe data layer<\/strong> uses a JSON-RPC 2.0-based exchange protocol that defines the message structure and semantics to handle what gets exchanged: connection initialization, connection termination between clients and servers, capability negotiation, and the actual functionality.<\/p>\n\n\n\n2. Transport layer<\/strong><\/h4>\n\n\n\nThe transport layer<\/strong> manages how information travels, either with stdio for local processes or HTTP for remote servers, handling security, authentication, and message delivery.<\/p>\n\n\n\nCore Components of MCP servers<\/strong><\/h3>\n\n\n\nMCP servers expose three key primitives that make AI actionable: <\/p>\n\n\n\n
\n- Tools<\/strong> are executable actions, such as querying databases or creating calendar events.<\/li>\n\n\n\n
- Resources<\/strong> provide context, like file content, database schemas, or API responses.<\/li>\n\n\n\n
- Prompts<\/strong> are reusable templates that structure AI interactions. <\/li>\n<\/ol>\n\n\n\n
![\"Agentic](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-8-2.png\")
<\/a><\/figure>\n\n\n\nWhy MCP Servers Matter for Agentic AI<\/strong><\/h3>\n\n\n\nAgentic AI systems need to act across multiple systems, data sources, and tools, and MCP<\/a> enables this with standardized discovery, data access, and action execution. It also keeps humans in control through sampling (i.e., servers can request AI completions) and elicitation (i.e., servers can ask for user input or confirmation).<\/p>\n\n\n\nSince you now know what an MCP server does and also why it matters for today\u2019s Agentic AI systems, let\u2019s explore how you can configure and use one yourself.<\/p>\n\n\n\n
How to set up an MCP Server<\/h2>\n\n\n\n
In this section, we\u2019ll walk through setting up a GitHub MCP server in VS Code so AI assistants like GitHub Copilot agent can have direct access to your GitHub account. <\/p>\n\n\n\n
Why GitHub? It\u2019s one of the most popular MCP servers and demonstrates the core concepts of the tech beautifully. Plus, the setup process is straightforward and mirrors how you\u2019d configure other MCP servers.<\/p>\n\n\n\n
Once you understand how to set up the GitHub MCP server, you\u2019ll have the foundation to connect AI to virtually any tool in your tech stack. <\/p>\n\n\n\n
Ready? Let’s get started!<\/p>\n\n\n\n
Prerequisite<\/strong><\/h3>\n\n\n\nFirst off, make sure that you\u2019ve got:<\/p>\n\n\n\n
\n- Visual Studio Code (version 1.101+) installed on your machine (tested with version 1.106.1).<\/li>\n\n\n\n
- A GitHub account. If you don\u2019t have an account, create one here<\/a>.<\/li>\n<\/ol>\n\n\n\n
Now that we\u2019ve got those out of the way, let\u2019s begin.<\/p>\n\n\n\n
Step 1: Create the MCP configuration structure<\/strong><\/h3>\n\n\n\nFirst, you need to create a special folder and file that VS Code will use to manage your MCP servers:<\/p>\n\n\n\n
\n- Launch VS Code and then choose a project folder where you want to use the GitHub MCP server.<\/li>\n\n\n\n
- In your project\u2019s root directory, create a new folder called .vscode (yes, with the dot at the beginning).<\/li>\n\n\n\n
- Inside the .vscode folder, create a file named mcp.json<\/li>\n<\/ol>\n\n\n\n
![\"Inside](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-7-2.png\")
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
![\"Select](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-11-1-1024x463.png\")
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
![\"Click](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-9-2-1024x266.png\")
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
![\"Click](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-12-1-1024x381.png\")
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
![\"New](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-14-2-1024x631.png\")
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
![\"Prompts](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-13-2-1024x729.png\")
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n![\"Atlassian](\"https:\/\/blog2workato.wpengine.com\/wp-content\/uploads\/2026\/10\/image-10-1.png\")
<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
Two layers power this:<\/p>\n\n\n\n
1. Data layer<\/strong><\/h4>\n\n\n\nThe data layer<\/strong> uses a JSON-RPC 2.0-based exchange protocol that defines the message structure and semantics to handle what gets exchanged: connection initialization, connection termination between clients and servers, capability negotiation, and the actual functionality.<\/p>\n\n\n\n2. Transport layer<\/strong><\/h4>\n\n\n\nThe transport layer<\/strong> manages how information travels, either with stdio for local processes or HTTP for remote servers, handling security, authentication, and message delivery.<\/p>\n\n\n\nCore Components of MCP servers<\/strong><\/h3>\n\n\n\nMCP servers expose three key primitives that make AI actionable: <\/p>\n\n\n\n
\n- Tools<\/strong> are executable actions, such as querying databases or creating calendar events.<\/li>\n\n\n\n
- Resources<\/strong> provide context, like file content, database schemas, or API responses.<\/li>\n\n\n\n
- Prompts<\/strong> are reusable templates that structure AI interactions. <\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nWhy MCP Servers Matter for Agentic AI<\/strong><\/h3>\n\n\n\nAgentic AI systems need to act across multiple systems, data sources, and tools, and MCP<\/a> enables this with standardized discovery, data access, and action execution. It also keeps humans in control through sampling (i.e., servers can request AI completions) and elicitation (i.e., servers can ask for user input or confirmation).<\/p>\n\n\n\nSince you now know what an MCP server does and also why it matters for today\u2019s Agentic AI systems, let\u2019s explore how you can configure and use one yourself.<\/p>\n\n\n\n
How to set up an MCP Server<\/h2>\n\n\n\n
In this section, we\u2019ll walk through setting up a GitHub MCP server in VS Code so AI assistants like GitHub Copilot agent can have direct access to your GitHub account. <\/p>\n\n\n\n
Why GitHub? It\u2019s one of the most popular MCP servers and demonstrates the core concepts of the tech beautifully. Plus, the setup process is straightforward and mirrors how you\u2019d configure other MCP servers.<\/p>\n\n\n\n
Once you understand how to set up the GitHub MCP server, you\u2019ll have the foundation to connect AI to virtually any tool in your tech stack. <\/p>\n\n\n\n
Ready? Let’s get started!<\/p>\n\n\n\n
Prerequisite<\/strong><\/h3>\n\n\n\nFirst off, make sure that you\u2019ve got:<\/p>\n\n\n\n
\n- Visual Studio Code (version 1.101+) installed on your machine (tested with version 1.106.1).<\/li>\n\n\n\n
- A GitHub account. If you don\u2019t have an account, create one here<\/a>.<\/li>\n<\/ol>\n\n\n\n
Now that we\u2019ve got those out of the way, let\u2019s begin.<\/p>\n\n\n\n
Step 1: Create the MCP configuration structure<\/strong><\/h3>\n\n\n\nFirst, you need to create a special folder and file that VS Code will use to manage your MCP servers:<\/p>\n\n\n\n
\n- Launch VS Code and then choose a project folder where you want to use the GitHub MCP server.<\/li>\n\n\n\n
- In your project\u2019s root directory, create a new folder called .vscode (yes, with the dot at the beginning).<\/li>\n\n\n\n
- Inside the .vscode folder, create a file named mcp.json<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
2. Transport layer<\/strong><\/h4>\n\n\n\nThe transport layer<\/strong> manages how information travels, either with stdio for local processes or HTTP for remote servers, handling security, authentication, and message delivery.<\/p>\n\n\n\nCore Components of MCP servers<\/strong><\/h3>\n\n\n\nMCP servers expose three key primitives that make AI actionable: <\/p>\n\n\n\n
\n- Tools<\/strong> are executable actions, such as querying databases or creating calendar events.<\/li>\n\n\n\n
- Resources<\/strong> provide context, like file content, database schemas, or API responses.<\/li>\n\n\n\n
- Prompts<\/strong> are reusable templates that structure AI interactions. <\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nWhy MCP Servers Matter for Agentic AI<\/strong><\/h3>\n\n\n\nAgentic AI systems need to act across multiple systems, data sources, and tools, and MCP<\/a> enables this with standardized discovery, data access, and action execution. It also keeps humans in control through sampling (i.e., servers can request AI completions) and elicitation (i.e., servers can ask for user input or confirmation).<\/p>\n\n\n\nSince you now know what an MCP server does and also why it matters for today\u2019s Agentic AI systems, let\u2019s explore how you can configure and use one yourself.<\/p>\n\n\n\n
How to set up an MCP Server<\/h2>\n\n\n\n
In this section, we\u2019ll walk through setting up a GitHub MCP server in VS Code so AI assistants like GitHub Copilot agent can have direct access to your GitHub account. <\/p>\n\n\n\n
Why GitHub? It\u2019s one of the most popular MCP servers and demonstrates the core concepts of the tech beautifully. Plus, the setup process is straightforward and mirrors how you\u2019d configure other MCP servers.<\/p>\n\n\n\n
Once you understand how to set up the GitHub MCP server, you\u2019ll have the foundation to connect AI to virtually any tool in your tech stack. <\/p>\n\n\n\n
Ready? Let’s get started!<\/p>\n\n\n\n
Prerequisite<\/strong><\/h3>\n\n\n\nFirst off, make sure that you\u2019ve got:<\/p>\n\n\n\n
\n- Visual Studio Code (version 1.101+) installed on your machine (tested with version 1.106.1).<\/li>\n\n\n\n
- A GitHub account. If you don\u2019t have an account, create one here<\/a>.<\/li>\n<\/ol>\n\n\n\n
Now that we\u2019ve got those out of the way, let\u2019s begin.<\/p>\n\n\n\n
Step 1: Create the MCP configuration structure<\/strong><\/h3>\n\n\n\nFirst, you need to create a special folder and file that VS Code will use to manage your MCP servers:<\/p>\n\n\n\n
\n- Launch VS Code and then choose a project folder where you want to use the GitHub MCP server.<\/li>\n\n\n\n
- In your project\u2019s root directory, create a new folder called .vscode (yes, with the dot at the beginning).<\/li>\n\n\n\n
- Inside the .vscode folder, create a file named mcp.json<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
Core Components of MCP servers<\/strong><\/h3>\n\n\n\nMCP servers expose three key primitives that make AI actionable: <\/p>\n\n\n\n
\n- Tools<\/strong> are executable actions, such as querying databases or creating calendar events.<\/li>\n\n\n\n
- Resources<\/strong> provide context, like file content, database schemas, or API responses.<\/li>\n\n\n\n
- Prompts<\/strong> are reusable templates that structure AI interactions. <\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nWhy MCP Servers Matter for Agentic AI<\/strong><\/h3>\n\n\n\nAgentic AI systems need to act across multiple systems, data sources, and tools, and MCP<\/a> enables this with standardized discovery, data access, and action execution. It also keeps humans in control through sampling (i.e., servers can request AI completions) and elicitation (i.e., servers can ask for user input or confirmation).<\/p>\n\n\n\nSince you now know what an MCP server does and also why it matters for today\u2019s Agentic AI systems, let\u2019s explore how you can configure and use one yourself.<\/p>\n\n\n\n
How to set up an MCP Server<\/h2>\n\n\n\n
In this section, we\u2019ll walk through setting up a GitHub MCP server in VS Code so AI assistants like GitHub Copilot agent can have direct access to your GitHub account. <\/p>\n\n\n\n
Why GitHub? It\u2019s one of the most popular MCP servers and demonstrates the core concepts of the tech beautifully. Plus, the setup process is straightforward and mirrors how you\u2019d configure other MCP servers.<\/p>\n\n\n\n
Once you understand how to set up the GitHub MCP server, you\u2019ll have the foundation to connect AI to virtually any tool in your tech stack. <\/p>\n\n\n\n
Ready? Let’s get started!<\/p>\n\n\n\n
Prerequisite<\/strong><\/h3>\n\n\n\nFirst off, make sure that you\u2019ve got:<\/p>\n\n\n\n
\n- Visual Studio Code (version 1.101+) installed on your machine (tested with version 1.106.1).<\/li>\n\n\n\n
- A GitHub account. If you don\u2019t have an account, create one here<\/a>.<\/li>\n<\/ol>\n\n\n\n
Now that we\u2019ve got those out of the way, let\u2019s begin.<\/p>\n\n\n\n
Step 1: Create the MCP configuration structure<\/strong><\/h3>\n\n\n\nFirst, you need to create a special folder and file that VS Code will use to manage your MCP servers:<\/p>\n\n\n\n
\n- Launch VS Code and then choose a project folder where you want to use the GitHub MCP server.<\/li>\n\n\n\n
- In your project\u2019s root directory, create a new folder called .vscode (yes, with the dot at the beginning).<\/li>\n\n\n\n
- Inside the .vscode folder, create a file named mcp.json<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
<\/a><\/figure>\n\n\n\nWhy MCP Servers Matter for Agentic AI<\/strong><\/h3>\n\n\n\nAgentic AI systems need to act across multiple systems, data sources, and tools, and MCP<\/a> enables this with standardized discovery, data access, and action execution. It also keeps humans in control through sampling (i.e., servers can request AI completions) and elicitation (i.e., servers can ask for user input or confirmation).<\/p>\n\n\n\nSince you now know what an MCP server does and also why it matters for today\u2019s Agentic AI systems, let\u2019s explore how you can configure and use one yourself.<\/p>\n\n\n\n
How to set up an MCP Server<\/h2>\n\n\n\n
In this section, we\u2019ll walk through setting up a GitHub MCP server in VS Code so AI assistants like GitHub Copilot agent can have direct access to your GitHub account. <\/p>\n\n\n\n
Why GitHub? It\u2019s one of the most popular MCP servers and demonstrates the core concepts of the tech beautifully. Plus, the setup process is straightforward and mirrors how you\u2019d configure other MCP servers.<\/p>\n\n\n\n
Once you understand how to set up the GitHub MCP server, you\u2019ll have the foundation to connect AI to virtually any tool in your tech stack. <\/p>\n\n\n\n
Ready? Let’s get started!<\/p>\n\n\n\n
Prerequisite<\/strong><\/h3>\n\n\n\nFirst off, make sure that you\u2019ve got:<\/p>\n\n\n\n
\n- Visual Studio Code (version 1.101+) installed on your machine (tested with version 1.106.1).<\/li>\n\n\n\n
- A GitHub account. If you don\u2019t have an account, create one here<\/a>.<\/li>\n<\/ol>\n\n\n\n
Now that we\u2019ve got those out of the way, let\u2019s begin.<\/p>\n\n\n\n
Step 1: Create the MCP configuration structure<\/strong><\/h3>\n\n\n\nFirst, you need to create a special folder and file that VS Code will use to manage your MCP servers:<\/p>\n\n\n\n
\n- Launch VS Code and then choose a project folder where you want to use the GitHub MCP server.<\/li>\n\n\n\n
- In your project\u2019s root directory, create a new folder called .vscode (yes, with the dot at the beginning).<\/li>\n\n\n\n
- Inside the .vscode folder, create a file named mcp.json<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
Since you now know what an MCP server does and also why it matters for today\u2019s Agentic AI systems, let\u2019s explore how you can configure and use one yourself.<\/p>\n\n\n\n
How to set up an MCP Server<\/h2>\n\n\n\n
In this section, we\u2019ll walk through setting up a GitHub MCP server in VS Code so AI assistants like GitHub Copilot agent can have direct access to your GitHub account. <\/p>\n\n\n\n
Why GitHub? It\u2019s one of the most popular MCP servers and demonstrates the core concepts of the tech beautifully. Plus, the setup process is straightforward and mirrors how you\u2019d configure other MCP servers.<\/p>\n\n\n\n
Once you understand how to set up the GitHub MCP server, you\u2019ll have the foundation to connect AI to virtually any tool in your tech stack. <\/p>\n\n\n\n
Ready? Let’s get started!<\/p>\n\n\n\n
Prerequisite<\/strong><\/h3>\n\n\n\nFirst off, make sure that you\u2019ve got:<\/p>\n\n\n\n
\n- Visual Studio Code (version 1.101+) installed on your machine (tested with version 1.106.1).<\/li>\n\n\n\n
- A GitHub account. If you don\u2019t have an account, create one here<\/a>.<\/li>\n<\/ol>\n\n\n\n
Now that we\u2019ve got those out of the way, let\u2019s begin.<\/p>\n\n\n\n
Step 1: Create the MCP configuration structure<\/strong><\/h3>\n\n\n\nFirst, you need to create a special folder and file that VS Code will use to manage your MCP servers:<\/p>\n\n\n\n
\n- Launch VS Code and then choose a project folder where you want to use the GitHub MCP server.<\/li>\n\n\n\n
- In your project\u2019s root directory, create a new folder called .vscode (yes, with the dot at the beginning).<\/li>\n\n\n\n
- Inside the .vscode folder, create a file named mcp.json<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
Now that we\u2019ve got those out of the way, let\u2019s begin.<\/p>\n\n\n\n
Step 1: Create the MCP configuration structure<\/strong><\/h3>\n\n\n\nFirst, you need to create a special folder and file that VS Code will use to manage your MCP servers:<\/p>\n\n\n\n
\n- Launch VS Code and then choose a project folder where you want to use the GitHub MCP server.<\/li>\n\n\n\n
- In your project\u2019s root directory, create a new folder called .vscode (yes, with the dot at the beginning).<\/li>\n\n\n\n
- Inside the .vscode folder, create a file named mcp.json<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
<\/a><\/figure>\n\n\n\nYour folder structure should now look like the image above.<\/p>\n\n\n\n
Step 2: Add the GitHub MCP Server.<\/strong><\/h3>\n\n\n\nNow comes the fun part! VS Code has a nice built-in way to add MCP servers:<\/p>\n\n\n\n
\n- Look for and click the \u201cAdd server\u201d<\/strong> button in VS Code (this should appear when you\u2019re on the mcp.json file).<\/li>\n\n\n\n
- When prompted, select \u201cNPM Package\u201d<\/strong> from the dropdown options.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\nTo allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
\n- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
\n- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n\n- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\nWe\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
\n- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\nStep 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
\n- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
\n- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\nIf everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\nIf this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\nIf you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\nServer not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
\n- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
\n- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
\n- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
\n- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\nHere\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
<\/a><\/figure>\n\n\n\n- \n
- A text field will appear asking for the package name. Type in exactly: @modelcontextprotocol\/server-github and press Enter<\/li>\n\n\n\n
- When the confirmation prompt appears, click \u201cAllow<\/strong>\u201d to proceed.<\/li>\n<\/ol>\n\n\n\n<\/a><\/figure>\n\n\n\n
- \n
- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n
After this, VS Code will automatically configure the basic setup for you and populate your mcp.json file with the necessary configuration.<\/p>\n\n\n\n
Step 3: Create your GitHub personal access token<\/strong><\/h3>\n\n\n\n
To allow the MCP server access to your GitHub data, you need to grant it permission. Here\u2019s how to do that:<\/p>\n\n\n\n
- \n
- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
<\/a><\/figure>\n\n\n\n- \n
- Give your token a descriptive name like \u201cVS Code MCP Server\u201d <\/strong>or any text that describes your aim. I use \u201ctestMCP\u201d<\/strong> for mine.<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
- \n
- Repo<\/strong> (full control of private repositories)<\/li>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- user<\/strong> (read user profile data)<\/li>\n<\/ul>\n\n\n\n
<\/a><\/figure>\n\n\n\n- \n
- Scroll down and click \u201cGenerate token\u201d<\/strong><\/li>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
Step 4: Add your token to the configuration<\/strong><\/h3>\n\n\n\n
We\u2019re almost there! Now, it’s time to connect everything:<\/p>\n\n\n\n
- \n
- Go back to VS Code and open your mcp.json file<\/li>\n\n\n\n
- You should see a placeholder that looks like \u201c<YOUR_TOKEN>\u201d<\/strong> or similar.<\/li>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
Your configuration should look something like this:<\/p>\n\n\n\n
{\n \"servers\": {\n \"github\": {\n \"command\": \"npx\",\n \"args\": [\n \"-y\",\n \"@modelcontextprotocol\/server-github\"\n ],\n \"env\": {\n \"GITHUB_PERSONAL_ACCESS_TOKEN\": \"<YOUR_TOKEN>\"\n },\n \"type\": \"stdio\"\n }\n },\n \"inputs\": []\n}<\/code><\/pre>\n\n\n\nFor security reasons, never commit this file with your real token to version control, and ensure to add <\/strong>.vscode\/mcp.json<\/strong> to your <\/strong>.gitignore<\/strong> file.<\/strong><\/p>\n\n\n\n
Step 5: Test your connection<\/h3>\n\n\n\n
You\u2019re all set! Now, let\u2019s make sure everything works:<\/p>\n\n\n\n
- \n
- Restart VS Code to ensure the MCP server initializes properly.<\/li>\n\n\n\n
- Open GitHub Copilot or your AI assistant in agent mode.<\/li>\n\n\n\n
- Try a prompt that requires GitHub access, like: <\/li>\n<\/ol>\n\n\n\n
- \n
- \u201cShow me my recent pull requests.\u201d<\/li>\n\n\n\n
- \u201cSummarize the latest commits.\u201d<\/li>\n<\/ul>\n\n\n\n<\/a><\/figure>\n\n\n\n
If everything is configured correctly, the AI assistant should be able to fetch and display information from your GitHub account like in the screenshot above.<\/p>\n\n\n\n
Common issues and troubleshooting tips<\/h2>\n\n\n\n
Here are some common hurdles you might run into during this process and how to fix them: <\/p>\n\n\n\n
1. Authentication failures <\/strong><\/h3>\n\n\n\n
If this occurs, your token likely has spaces or is incomplete. Classic tokens start with gph_, so check and copy it fresh from GitHub without any extra characters.<\/p>\n\n\n\n
2. Access denied errors<\/strong><\/h3>\n\n\n\n
If you see this, you didn\u2019t grant enough permissions. Go back to GitHub token settings and enable at least the repo scope for repository access.<\/p>\n\n\n\n
3. Server won\u2019t start<\/strong><\/h3>\n\n\n\n
Server not starting? Check that your file is named exactly mcp.json (not mcp.json.txt) and it is directly inside .vscode. Also, scan for JSON syntax errors; missing commas and brackets are common culprits.<\/p>\n\n\n\n
Scaling AI across the organization with Workato Enterprise MCP<\/h2>\n\n\n\n
The GitHub MCP server works great for developers. But what if you want to connect AI across your entire business? <\/p>\n\n\n\n
Workato Enterprise MCP<\/a> extends MCP beyond single tools, letting AI assistants like Claude securely access multiple systems with enterprise-grade security<\/a>, governance, and human oversight.<\/p>\n\n\n\n
<\/a><\/figure>\n\n\n\nReal-world use cases<\/h3>\n\n\n\n
Here are some real examples of how Workato Enterprise MCP helps teams work faster with AI:<\/p>\n\n\n\n
- \n
- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
- \n
- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
- \n
- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
- \n
- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
<\/li>\n<\/ol>\n\n\n\nAs Alan Braun (VP Product Ecosystem, Atlassian) notes: \u201cAtlassian Rovo MCP Server and Workato’s Enterprise MCP together allow customers to unlock autonomous actions across Jira and Confluence with the security, governance, and skills required for AI at scale.\u201d<\/p>\n\n\n\n
Explore hundreds of pre-built, secure enterprise MCP servers at Workato<\/a> and see how easily you can create your own to power AI across your business.<\/p>\n\n\n\n
Here\u2019s to unlocking the full promise of agentic AI!<\/p>\n\n\n\n
- Marketers often can\u2019t link campaign spending to revenue. Workato Enterprise MCP lets AI combine CRM, automation<\/a>, and ad data to connect campaigns to the pipeline, summarize results, and suggest smarter budgets to boost ROI.
- Product teams can also ship faster, since it enables AI assistants to check GitHub commits, Jira, and other tools simultaneously and point out risks while still allowing oversight from the team.<\/li>\n<\/ol>\n\n\n\n
- With Workato MCP, AI assistants can deliver daily briefings to leaders that unify KPIs from sales, finance, HR, and ops with market news for real-time governed insights.<\/li>\n<\/ol>\n\n\n\n
- Workato Enterprise MCP helps sales teams create better outreach faster. With it, AI pulls insights from emails, summarizes Gong calls, and researches customer sites, thereby turning hours of work into seconds, with reps approving every action before it\u2019s carried out.<\/li>\n<\/ol>\n\n\n\n
- Replace the entire placeholder (including the angle brackets) with your actual GitHub token from step 3.<\/li>\n<\/ol>\n\n\n\n
- Important<\/strong>: Once the token is generated, copy it immediately! GitHub only shows it once, and you\u2019ll need it in the next step.<\/li>\n<\/ol>\n\n\n\n
- Read:org<\/strong> (if you work with organization repos)<\/li>\n\n\n\n
- Select the scopes\/permissions your server needs. For most use cases, you\u2019ll want:<\/li>\n<\/ol>\n\n\n\n
- Click the \u201cGenerate new token\u201d <\/strong>button (preferably, choose \u201cGenerate new token (classic)\u201d<\/strong>).<\/li>\n<\/ol>\n\n\n\n
- Log in to your GitHub account and go to the token creation page<\/a>.<\/li>\n\n\n\n
- Next, you will be prompted for your GitHub Access Token. Press \u201cEscape<\/strong>\u201d to cancel. We\u2019ll configure this manually in the next step.<\/li>\n<\/ol>\n\n\n\n