Guide

MCP Integration: Turn Your Team's Knowledge Into Portable Context for Any AI Tool

Learn how to use Zine as both an MCP server (share your knowledge with Cursor, VS Code, Claude) and MCP client (connect Sentry, GitHub, and other tools).

The Model Context Protocol (MCP), developed by Anthropic, is revolutionizing how AI tools share context. Instead of every AI tool having its own proprietary way to access data, MCP provides an open standard that works across Cursor, VS Code, Claude Desktop, and more.

Zine takes MCP to the next level by acting as both server and client:

  • Zine as MCP Server: Share your team's knowledge (Slack, GitHub, email, docs) with any MCP-compatible AI tool
  • Zine as MCP Client: Connect other MCP servers (Sentry, filesystem, databases) to Zine chat

This guide shows you how to set up both, with practical examples that will transform your development workflow.


Table of Contents

  1. What is MCP?
  2. Why MCP Matters for Developers
  3. Zine as MCP Server: Setup
  4. Using Zine in Cursor
  5. Using Zine in VS Code
  6. Using Zine in Windsurf
  7. Using Zine in Claude Code
  8. Zine as MCP Client: Connecting Other Tools
  9. Practical Workflows
  10. Troubleshooting

What is MCP?

The Model Context Protocol (MCP) is an open standard created by Anthropic for connecting AI applications to data sources. Think of it as "USB for AI context"—one protocol that works everywhere.

Before MCP:

  • Cursor has custom integrations for specific tools
  • VS Code Copilot has its own connector system
  • Claude Desktop can't access your company data
  • Every tool requires custom development to add new sources

With MCP:

  • Write one MCP server → works in Cursor, VS Code, Claude Desktop, Windsurf, and any MCP-compatible tool
  • Your coding agent can access company Slack, GitHub, docs through standardized protocol
  • Tool vendors don't need to build every integration—they support MCP, users bring their own servers

How MCP Works:

┌─────────────────┐
│  AI Tool        │  (Cursor, VS Code, Claude Desktop)
│  (MCP Client)   │
└────────┬────────┘
         │ MCP Protocol
         │ (standardized)
┌────────▼────────┐
│  MCP Server     │  (Zine, Sentry, GitHub, Filesystem)
│  (Data Source)  │
└─────────────────┘

MCP Server: Exposes data/tools to AI applications (e.g., Zine exposes your Slack/GitHub/docs) MCP Client: AI application that consumes data from MCP servers (e.g., Cursor)

Zine is unique: It's both a server (share your knowledge) and a client (connect other servers).


Why MCP Matters for Developers

The Problem: Generic Coding Agents

Without context:

  • Developer: "How does our authentication work?"
  • Cursor (no context): Gives generic OAuth tutorial from web training
  • Reality: Your team uses custom auth service with Postgres sessions

You spend 10 minutes: Finding Slack thread, reading GitHub code, checking Notion spec

The Solution: Context-Aware Agents with MCP

With Zine MCP:

  • Developer: "How does our authentication work?"
  • Cursor queries Zine MCP
  • Zine searches: Slack + GitHub + Notion + meeting recordings
  • Cursor responds: "Your team uses a custom auth service (auth-service repo) with Postgres sessions. See the Notion spec 'Auth Architecture v2' and the Slack discussion where JWT was rejected."

Time saved: 10 minutes → 30 seconds

Context quality: Generic web knowledge → Your team's specific implementation

Real Developer Workflows Transformed

Scenario 1: Implementing a New Feature

  • Task: "Add password reset flow"
  • Without MCP: Google similar implementations, hope you follow team patterns
  • With Zine MCP: "How did we implement email verification?" → Zine returns your team's implementation (GitHub PR #234) + security considerations discussed in Slack + Notion security checklist

Scenario 2: Debugging Production Issues

  • Error: Checkout API timeout spike
  • Without MCP: Check Sentry, search Slack #incidents, look for recent GitHub changes manually
  • With Zine MCP + Sentry MCP: "What caused checkout timeouts?" → Zine finds Slack incident discussion + GitHub PR that added Redis cache + Sentry error traces

Scenario 3: Code Review

  • PR: New payment processing logic
  • Without MCP: Review code, hope you know team standards
  • With Zine MCP: "Check this PR against our security guidelines" → Zine surfaces team's security checklist from Notion + past PR review comments on similar code

Zine as MCP Server: Setup

Prerequisites

  1. Zine Account: Sign up at zine.ai
  2. Connected Data: Add Slack, GitHub, Gmail, Notion, or other sources to Zine
  3. MCP-Compatible Tool: Cursor, VS Code, Windsurf, Claude Desktop, or Claude Code

Get Your Zine MCP URL

Remote MCP Server (easiest):

Zine hosts the MCP server for you. No local setup required.

URL: https://www.zine.ai/mcp
API Key: (get from Zine settings)

To get your API key:

  1. Log in to Zine
  2. Go to Settings → API Keys
  3. Create a new key (or copy existing)
  4. Label it "MCP Integration"

Keep this key secure — it grants access to your Zine workspace.


Using Zine in Cursor

Cursor is the leading AI-first code editor. Adding Zine MCP gives Cursor access to your team's knowledge.

Step 1: Configure Cursor MCP

Mac/Linux:

# Edit Cursor MCP config
nano ~/.cursor/mcp_servers.json

Windows:

# Edit Cursor MCP config
notepad %USERPROFILE%\.cursor\mcp_servers.json

Step 2: Add Zine Server

{
  "mcpServers": {
    "zine": {
      "type": "sse",
      "url": "https://www.zine.ai/mcp",
      "apiKey": "your-zine-api-key-here"
    }
  }
}

Replace your-zine-api-key-here with your actual Zine API key.

Step 3: Restart Cursor

  • Quit Cursor completely
  • Reopen Cursor
  • MCP connection will be established automatically

Step 4: Test the Integration

Open Cursor chat (Cmd+L or Ctrl+L) and try:

Search Zine for our authentication implementation

If configured correctly, Cursor will:

  1. Query Zine MCP
  2. Zine searches your Slack, GitHub, Notion
  3. Cursor displays results with citations

Example Queries in Cursor with Zine

Architecture Questions:

Search Zine for database choice rationale
How did we decide on Postgres vs MongoDB?

Implementation Patterns:

Search Zine for API rate limiting implementation
How does our team handle caching?

Recent Discussions:

Search Zine for recent Slack discussions about performance
What did the team say about Redis in the last 30 days?

Bug Context:

Search Zine for checkout API timeout incidents
Has this error happened before?

Team Conventions:

Search Zine for code review guidelines
What are our testing standards?

Video Tutorial

Watch: Setting up Zine MCP with Cursor


Using Zine in VS Code

VS Code with GitHub Copilot now supports MCP (as of 2025).

Step 1: Install Copilot Chat Extension

Ensure you have GitHub Copilot Chat extension installed and active.

Step 2: Configure VS Code MCP

Mac/Linux:

# Edit VS Code MCP config
nano ~/.vscode/mcp_servers.json

Windows:

# Edit VS Code MCP config
notepad %USERPROFILE%\.vscode\mcp_servers.json

Step 3: Add Zine Server

{
  "mcpServers": {
    "zine": {
      "type": "sse",
      "url": "https://www.zine.ai/mcp",
      "apiKey": "your-zine-api-key-here"
    }
  }
}

Step 4: Restart VS Code

Close and reopen VS Code for changes to take effect.

Step 5: Use in Copilot Chat

Open Copilot Chat panel and query:

@zine Search for authentication implementation
@zine How does our payment processing work?

Note: Prefix with @zine to explicitly query the Zine MCP server.


Using Zine in Windsurf

Windsurf by Codeium supports MCP through its Cascade AI feature.

Step 1: Configure Windsurf MCP

Mac/Linux:

# Edit Windsurf MCP config
nano ~/.windsurf/mcp_servers.json

Windows:

# Edit Windsurf MCP config
notepad %APPDATA%\Windsurf\mcp_servers.json

Step 2: Add Zine Server

{
  "mcpServers": {
    "zine": {
      "type": "sse",
      "url": "https://www.zine.ai/mcp",
      "apiKey": "your-zine-api-key-here"
    }
  }
}

Step 3: Restart Windsurf

Quit and reopen Windsurf.

Step 4: Use in Cascade AI

In Cascade chat, query Zine:

Search Zine for our API architecture decisions
@zine What did Alice say about Redis?

Using Zine in Claude Code

Claude Code is Anthropic's coding assistant (browser-based).

Step 1: Open Claude Code Settings

Navigate to Settings → MCP Servers

Step 2: Add Zine Server

Click "Add Server" and enter:

Name: Zine
Type: Remote (SSE)
URL: https://www.zine.ai/mcp
API Key: your-zine-api-key-here

Step 3: Save and Test

Ask Claude Code:

Search my Zine workspace for authentication discussion

Claude will query Zine MCP and return results.


Zine as MCP Client: Connecting Other Tools

Zine doesn't just expose your knowledge—it can also consume other MCP servers. This lets you combine multiple tools in one search.

Available MCP Servers

Popular MCP servers you can connect to Zine:

  1. Sentry MCP: Query error traces and issues
  2. GitHub MCP: Search repos (even if GitHub connector isn't set up)
  3. Filesystem MCP: Access local files on your machine
  4. Database MCP: Query databases directly
  5. Slack MCP: Alternative Slack access (if using Slack's official MCP)

Example: Connecting Sentry MCP to Zine

Step 1: Add Sentry MCP Server in Zine

  1. Open Zine → Settings → MCP Servers
  2. Click "Add MCP Server"
  3. Enter:
    Name: Sentry
    Type: Remote
    URL: (Sentry MCP endpoint)
    API Key: (Your Sentry API key)
    

Step 2: Use in Zine Chat

Now when you ask Zine:

What Sentry errors are related to checkout API?

Zine will:

  1. Query Sentry MCP for error traces
  2. Search Zine's own knowledge (Slack #incidents, GitHub recent changes)
  3. Synthesize: "Sentry shows 142 errors from checkout-service. Slack #incidents has a thread from 2 hours ago. GitHub PR #567 was merged yesterday adding Redis cache."

Result: Unified incident context from multiple tools in one query.


Practical Workflows

Workflow 1: New Feature Implementation

Task: Implement password reset flow

Step 1: Research Existing Patterns

In Cursor:

Search Zine for email verification implementation

Zine Returns:

  • GitHub PR #234: "Add email verification"
  • Notion doc: "Email Security Guidelines"
  • Slack #engineering: Thread discussing verification UX

Step 2: Check Security Requirements

Search Zine for password security requirements

Zine Returns:

  • Notion: "Security Checklist"
  • Slack: Discussion about password hashing (bcrypt selected)
  • GitHub: Past PR reviews mentioning password handling

Step 3: Implement with Context

Write code, Cursor has context from Zine:

  • Follows team's email verification pattern
  • Uses approved security libraries (bcrypt)
  • Matches team's API response format

Time Saved: 30-60 minutes of manual searching


Workflow 2: Incident Response

Scenario: Checkout API throwing 500 errors

Step 1: Query Multi-Tool Context

In Zine (with Sentry MCP connected):

Checkout API errors in the last hour

Zine Returns:

  • Sentry MCP: 87 errors, stack trace shows Redis timeout
  • Slack #incidents: No current discussion (new issue)
  • GitHub: PR #567 merged yesterday: "Improve Redis caching"

Hypothesis: Recent Redis change caused the issue

Step 2: Find Related Discussions

Search Zine for Redis cache discussions from last week

Zine Returns:

  • Slack thread: "Concerns about Redis eviction policy raised by Bob"
  • Meeting recording: Architecture review (timestamp: 23:42 - cache timeout discussion)

Step 3: Quick Fix

You now know:

  • The recent change (GitHub PR #567)
  • The team's concerns (Slack discussion)
  • How to roll back or adjust (meeting context)

Time Saved: 15-30 minutes of context gathering


Workflow 3: Code Review with Full Context

Scenario: Reviewing a PR that adds new API endpoint

Step 1: Understand Current API Design

In Cursor:

Search Zine for API design patterns

Zine Returns:

  • Notion: "API Design Guide"
  • GitHub: Example PRs for similar endpoints
  • Slack: Team discussion about REST vs GraphQL decisions

Step 2: Check Security Considerations

Search Zine for API security guidelines

Zine Returns:

  • Notion: "API Security Checklist" (rate limiting, auth, input validation)
  • Past PR comments: "Always validate user input in endpoint handlers"

Step 3: Review PR

With context from Zine, you can:

  • Confirm PR follows team patterns
  • Check security guidelines are met
  • Reference past discussions if concerns arise

Time Saved: 10-20 minutes per review


Troubleshooting

Connection Issues

Problem: Cursor/VS Code not connecting to Zine MCP

Solutions:

  1. Check API Key: Verify it's correct in config file
  2. Restart IDE: Fully quit and reopen
  3. Check Zine Status: Visit zine.ai to ensure service is up
  4. Network: Ensure no corporate firewall blocking zine.ai

Test Connection:

# Test Zine MCP endpoint (Mac/Linux)
curl -H "Authorization: Bearer your-api-key" https://www.zine.ai/mcp/health

Empty Results

Problem: Queries return no results

Possible Causes:

  1. No Data Connected: Ensure you've connected Slack, GitHub, etc. in Zine
  2. Data Not Synced: Initial sync can take 30min - 2 hours
  3. Query Too Specific: Try broader search terms
  4. Permissions: Ensure your Zine account has access to the sources

Solution: Check Zine dashboard → Data Sources → verify sync status


Slow Responses

Problem: MCP queries take 10+ seconds

Causes:

  1. Large Dataset: First query can be slow (subsequent queries cached)
  2. Complex Search: Cross-tool queries with many results
  3. Network Latency: Remote MCP adds network overhead

Solutions:

  • Be specific in queries (reduces search scope)
  • Use saved views in Zine (pre-filtered searches)
  • Consider upgrading Zine plan (higher performance limits)

API Key Errors

Problem: "Unauthorized" or "Invalid API key"

Solutions:

  1. Regenerate Key: Zine Settings → API Keys → Create new key
  2. Check Quotes: Ensure API key in JSON has no extra quotes/spaces
  3. Update Config: Make sure config file saved correctly
  4. Restart IDE: Changes may not apply until restart

Next Steps

Now that you have MCP set up:

  1. Connect Your Data: Add Slack, GitHub, Gmail, Notion to Zine
  2. Try Example Queries: Use the example queries in this guide
  3. Create Saved Views: Pre-filter for common searches ("Security Guidelines", "API Patterns")
  4. Add More MCP Servers: Connect Sentry, filesystem, or other tools to Zine
  5. Share with Team: Help teammates set up MCP for collective benefit

Related Guides:

Learn More:


MCP turns your team's knowledge into portable context for any AI tool. Set it up once, use it everywhere.

Ready to Build with Graphlit?

Start building AI-powered applications with our API-first platform. Free tier includes 100 credits/month — no credit card required.

No credit card required • 5 minutes to first API call

MCP Integration: Turn Your Team's Knowledge Into Portable Context for Any AI Tool | Graphlit Developer Guides