MCP Servers

Model Context Protocol (MCP) servers extend ClaudeAutoPM with additional capabilities. This guide covers the integration of Context7, Playwright, filesystem servers, and other MCP tools.

What is MCP?

The Model Context Protocol (MCP) is a standardized way for AI assistants to communicate with external tools and data sources. MCP servers provide:

Live documentation - Current API docs, not training data
External tools - Browser automation, file operations
Data access - Databases, cloud services, APIs
Context sharing - Between agents and sessions

Available MCP Servers

Context7

Context7 provides live documentation access for current best practices.

Purpose: Query up-to-date documentation instead of relying on training data.

Why Use Context7:

Training data becomes stale over time
APIs change between versions
Best practices evolve
Breaking changes are not reflected in older models

Usage:

bash

# Query documentation
mcp://context7/react/hooks

# Technology-specific docs
mcp://context7/fastapi/endpoints
mcp://context7/kubernetes/deployments
mcp://context7/aws/lambda

Integration with Commands:

Every command in ClaudeAutoPM includes required Context7 queries:

markdown

**Documentation Queries:**
- `mcp://context7/agile/epic-decomposition` - Epic breakdown best practices
- `mcp://context7/agile/task-sizing` - Task estimation guidance

Setup:

bash

# Configure Context7 integration
/mcp:context-setup --server=context7

Playwright MCP

Browser automation and testing capabilities.

Purpose: Automate browser interactions for testing and scraping.

Capabilities:

Navigate web pages
Fill forms and click elements
Take screenshots
Execute JavaScript
Handle authentication flows

Usage Examples:

bash

# Scaffold Playwright tests for a feature
/playwright:test-scaffold user-authentication

# Use with e2e-test-engineer agent
@e2e-test-engineer create Playwright tests for the checkout flow

Test Generation:

javascript

// Generated test example
import { test, expect } from '@playwright/test';

test('user can complete checkout', async ({ page }) => {
  await page.goto('/cart');
  await page.click('[data-testid="checkout-button"]');
  await page.fill('#email', 'user@example.com');
  await page.click('[data-testid="submit"]');
  await expect(page).toHaveURL('/confirmation');
});

Filesystem MCP

Enhanced file operations with validation.

Purpose: Structured file operations with safety checks.

Capabilities:

Read files with validation
Write with backup creation
Atomic file operations
Directory traversal safety

Usage:

bash

# Configure filesystem server
/mcp:context-setup --server=filesystem

Custom MCP Servers

You can integrate additional MCP servers for specific needs.

Adding a Server:

bash

# Find available servers
mcp-find "database"

# Add a server
mcp-add postgres-mcp

# Configure the server
mcp-config-set postgres-mcp {
  "host": "localhost",
  "port": 5432,
  "database": "myapp"
}

MCP Configuration

Setup Commands

bash

# Set up MCP context sharing
/mcp:context-setup --server=context7 --pool-name=project-context

# Configure specific agents for MCP
/mcp:context-setup --agents=python-backend-engineer,react-ui-expert

# Set context size limits
/mcp:context-setup --max-size=50MB

Documentation Refresh

Keep documentation caches current:

bash

# Refresh all documentation
/mcp:docs-refresh

# Refresh for specific technology
/mcp:docs-refresh --tech=fastapi

# Force refresh ignoring cache
/mcp:docs-refresh --force

# Validate documentation integrity
/mcp:docs-refresh --validate

Context Pool Management

Share context between agents:

bash

# Create a shared context pool
/mcp:context-setup --pool-name=feature-auth --agents=all

# Update pool contents
/context:update

# Prime context for new session
/context:prime

Context7 Deep Dive

Querying Documentation

Context7 queries follow this pattern:

mcp://context7/<technology>/<topic>

Common Queries:

Technology	Query Example	What You Get
React	`mcp://context7/react/hooks`	Current hooks API
FastAPI	`mcp://context7/fastapi/dependencies`	Dependency injection docs
AWS	`mcp://context7/aws/lambda`	Lambda best practices
Kubernetes	`mcp://context7/kubernetes/pods`	Pod configuration
PostgreSQL	`mcp://context7/postgresql/indexes`	Indexing strategies

Agent Integration

Agents automatically query Context7 before implementation:

bash

@python-backend-expert create a FastAPI endpoint

# Agent queries:
# mcp://context7/fastapi/routes
# mcp://context7/fastapi/dependencies
# mcp://context7/pydantic/models

Enforcement

ClaudeAutoPM enforces Context7 usage:

Pre-command hooks verify required queries
Pre-agent hooks ensure documentation is queried
Validation rules block implementations without Context7

You will see feedback like:

Context7 Enforcement Active

Command: /pm:epic-decompose
Querying Context7 for required documentation...

   -> mcp://context7/agile/epic-decomposition
   -> mcp://context7/agile/task-sizing

Context7 queries complete
Key findings: INVEST criteria for stories, 1-3 day task sizing

Proceeding with epic decomposition...

Playwright Deep Dive

Test Scaffolding

Generate test structure for a feature:

bash

/playwright:test-scaffold user-dashboard

This creates:

tests/
  e2e/
    user-dashboard/
      dashboard.spec.ts
      fixtures/
        dashboard.fixture.ts
      pages/
        dashboard.page.ts

Page Object Pattern

Generated page objects:

typescript

// pages/dashboard.page.ts
import { Page, Locator } from '@playwright/test';

export class DashboardPage {
  readonly page: Page;
  readonly welcomeMessage: Locator;
  readonly userMenu: Locator;

  constructor(page: Page) {
    this.page = page;
    this.welcomeMessage = page.getByTestId('welcome');
    this.userMenu = page.getByRole('button', { name: 'User menu' });
  }

  async goto() {
    await this.page.goto('/dashboard');
  }

  async openUserMenu() {
    await this.userMenu.click();
  }
}

Test Fixtures

Reusable test setup:

typescript

// fixtures/dashboard.fixture.ts
import { test as base } from '@playwright/test';
import { DashboardPage } from '../pages/dashboard.page';

type Fixtures = {
  dashboardPage: DashboardPage;
};

export const test = base.extend<Fixtures>({
  dashboardPage: async ({ page }, use) => {
    const dashboardPage = new DashboardPage(page);
    await dashboardPage.goto();
    await use(dashboardPage);
  },
});

Running Tests

bash

# Run all Playwright tests
npx playwright test

# Run specific test file
npx playwright test tests/e2e/user-dashboard/

# Run with UI mode
npx playwright test --ui

# Generate test report
npx playwright show-report

Custom Configuration

MCP Server Configuration

Create a configuration file at .claude/mcp-config.json:

json

{
  "servers": {
    "context7": {
      "enabled": true,
      "cache_ttl": 3600,
      "priority": "high"
    },
    "playwright": {
      "enabled": true,
      "browser": "chromium",
      "headless": true
    },
    "filesystem": {
      "enabled": true,
      "root": ".",
      "allowed_extensions": [".js", ".ts", ".py", ".md"]
    }
  },
  "context_pools": {
    "default": {
      "max_size": "100MB",
      "agents": ["all"]
    },
    "testing": {
      "max_size": "50MB",
      "agents": ["test-runner", "e2e-test-engineer"]
    }
  }
}

Environment Variables

bash

# Context7 configuration
export CONTEXT7_API_KEY="your-api-key"
export CONTEXT7_CACHE_DIR=".claude/context7-cache"

# Playwright configuration
export PLAYWRIGHT_BROWSERS_PATH=".playwright"
export PLAYWRIGHT_SKIP_BROWSER_DOWNLOAD=0

MCP Manager Agent

The @mcp-manager agent helps configure MCP integrations:

bash

# Set up MCP for project
@mcp-manager configure Context7 for this Python/FastAPI project

# Troubleshoot MCP issues
@mcp-manager diagnose Context7 connection issues

# Optimize MCP performance
@mcp-manager optimize context pool configuration

Best Practices

Context7 Usage

Always query before implementing - Never rely solely on training data
Query specific topics - More specific queries get better results
Verify against current docs - Especially for API signatures
Update caches regularly - Use /mcp:docs-refresh periodically

Playwright Integration

Use page objects - Keep tests maintainable
Create fixtures - Reuse common setup
Test critical paths - Focus on user journeys
Run in CI - Automate test execution

Performance Optimization

Enable caching - Reduce repeated queries
Set appropriate pool sizes - Balance memory and performance
Use specific servers - Only enable what you need
Monitor context usage - Track agent efficiency

Troubleshooting

Context7 Issues

Connection Failed:

bash

# Check API key
echo $CONTEXT7_API_KEY

# Verify network access
curl https://api.context7.dev/health

# Clear cache and retry
rm -rf .claude/context7-cache
/mcp:docs-refresh --force

Stale Documentation:

bash

# Force refresh
/mcp:docs-refresh --tech=fastapi --force

# Validate cache
/mcp:docs-refresh --validate

Playwright Issues

Browser Not Found:

bash

# Install browsers
npx playwright install

# Install specific browser
npx playwright install chromium

Test Timeout:

typescript

// Increase timeout in config
export default defineConfig({
  timeout: 60000,
  expect: {
    timeout: 10000,
  },
});

General MCP Issues

Server Not Available:

bash

# Check server status
mcp-status context7

# Restart server
mcp-restart context7

# Check logs
mcp-logs context7

Next Steps

Review Best Practices for optimal MCP usage
Explore the Agent Registry for MCP-enabled agents
Check Configuration Reference for detailed options

MCP Servers ​

What is MCP? ​

Available MCP Servers ​

Context7 ​

Playwright MCP ​

Filesystem MCP ​

Custom MCP Servers ​

MCP Configuration ​

Setup Commands ​

Documentation Refresh ​

Context Pool Management ​

Context7 Deep Dive ​

Querying Documentation ​

Agent Integration ​

Enforcement ​

Playwright Deep Dive ​

Test Scaffolding ​

Page Object Pattern ​

Test Fixtures ​

Running Tests ​

Custom Configuration ​

MCP Server Configuration ​

Environment Variables ​

MCP Manager Agent ​

Best Practices ​

Context7 Usage ​

Playwright Integration ​

Performance Optimization ​

Troubleshooting ​

Context7 Issues ​

Playwright Issues ​

General MCP Issues ​

Next Steps ​

MCP Servers

What is MCP?

Available MCP Servers

Context7

Playwright MCP

Filesystem MCP

Custom MCP Servers

MCP Configuration

Setup Commands

Documentation Refresh

Context Pool Management

Context7 Deep Dive

Querying Documentation

Agent Integration

Enforcement

Playwright Deep Dive

Test Scaffolding

Page Object Pattern

Test Fixtures

Running Tests

Custom Configuration

MCP Server Configuration

Environment Variables

MCP Manager Agent

Best Practices

Context7 Usage

Playwright Integration

Performance Optimization

Troubleshooting

Context7 Issues

Playwright Issues

General MCP Issues

Next Steps