Skip to main content
Glama

Personal Resume Agent

by vsiwach
DEPLOYMENT.md4.5 kB
# Personal Resume MCP Server - AWS Deployment Guide This guide walks you through deploying the Personal Resume MCP Server to AWS for remote access via Claude Desktop. ## Architecture Overview ``` Claude Desktop → HTTP Request → ALB → ECS Fargate → MCP Server → Resume RAG System ``` ## Prerequisites 1. AWS CLI configured with appropriate permissions 2. Docker installed and running 3. Your resume files in the `data/` directory ## Quick Start (Local Testing) Test the HTTP server locally before deploying: ```bash # Build and run locally docker-compose up --build # Test the server curl http://localhost:8000/health curl -X POST http://localhost:8000/mcp \ -H "Content-Type: application/json" \ -d '{"jsonrpc": "2.0", "id": 1, "method": "tools/list"}' ``` ## AWS Deployment Options ### Option 1: Automated CloudFormation Deployment 1. **Set environment variables:** ```bash export AWS_REGION=us-east-1 export AWS_ACCOUNT_ID=$(aws sts get-caller-identity --query Account --output text) ``` 2. **Deploy infrastructure:** ```bash # Build and push image cd deploy ./deploy.sh # Deploy infrastructure aws cloudformation create-stack \ --stack-name personal-resume-mcp-stack \ --template-body file://cloudformation-template.yaml \ --parameters ParameterKey=ImageURI,ParameterValue=$AWS_ACCOUNT_ID.dkr.ecr.$AWS_REGION.amazonaws.com/personal-resume-mcp:latest \ --capabilities CAPABILITY_NAMED_IAM \ --region $AWS_REGION ``` 3. **Get the endpoint:** ```bash aws cloudformation describe-stacks \ --stack-name personal-resume-mcp-stack \ --query 'Stacks[0].Outputs[?OutputKey==`MCPEndpoint`].OutputValue' \ --output text ``` ### Option 2: Manual ECS Deployment 1. **Build and push image:** ```bash cd deploy ./deploy.sh ``` 2. **Create ECS cluster and service using AWS Console or CLI** ## Claude Desktop Configuration Once deployed, update your Claude Desktop configuration: ```json { "mcpServers": { "personal-resume": { "command": "curl", "args": [ "-X", "POST", "http://YOUR_ALB_ENDPOINT/mcp", "-H", "Content-Type: application/json", "-d", "@-" ] } } } ``` ## Testing the Deployment 1. **Health check:** ```bash curl http://YOUR_ALB_ENDPOINT/health ``` 2. **Test MCP tools:** ```bash # List available tools curl -X POST http://YOUR_ALB_ENDPOINT/mcp \ -H "Content-Type: application/json" \ -d '{"jsonrpc": "2.0", "id": 1, "method": "tools/list"}' # Query resume curl -X POST http://YOUR_ALB_ENDPOINT/mcp \ -H "Content-Type: application/json" \ -d '{ "jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": { "name": "query_resume", "arguments": { "query": "What programming languages do I know?" } } }' ``` ## Available MCP Tools - `query_resume`: Ask questions about resume content - `get_agent_info`: Get agent capabilities and status - `analyze_skill_match`: Compare skills with job requirements ## Security Considerations - The ALB is publicly accessible (required for Claude Desktop) - Consider adding authentication/authorization if needed - Resume data is processed locally within the container - No external API calls for embeddings (uses local sentence-transformers) ## Monitoring and Logs - CloudWatch logs: `/ecs/personal-resume-mcp-server` - Health check endpoint: `/health` - Metrics available in ECS console ## Cost Estimation Approximate monthly costs (us-east-1): - ECS Fargate (1 task, 1 vCPU, 2GB): ~$30 - Application Load Balancer: ~$18 - CloudWatch Logs (minimal): ~$1 - **Total: ~$50/month** ## Troubleshooting ### Common Issues: 1. **Container fails to start:** - Check CloudWatch logs - Verify resume files are present in container 2. **Health check failing:** - Ensure port 8000 is exposed - Check security group rules 3. **Claude Desktop can't connect:** - Verify ALB endpoint is accessible - Check CORS settings in the HTTP server ### Debug Commands: ```bash # Check ECS service status aws ecs describe-services --cluster personal-resume-mcp-cluster --services personal-resume-mcp-service # View logs aws logs tail /ecs/personal-resume-mcp-server --follow # Test connectivity curl -v http://YOUR_ALB_ENDPOINT/health ``` ## Cleanup To avoid ongoing charges: ```bash # Delete CloudFormation stack aws cloudformation delete-stack --stack-name personal-resume-mcp-stack # Delete ECR repository aws ecr delete-repository --repository-name personal-resume-mcp --force ```

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/vsiwach/MCP-Resume-AWS'

If you have feedback or need assistance with the MCP directory API, please join our Discord server