vercel -for-zed

The AWS MCP Server for Zed provides a local MCP interface that lets Zed's AI Assistant query and interact with AWS services from within the editor. Once running, the server can access AWS resources such as S3, DynamoDB, and Lambda, enabling the AI to retrieve information, perform lookups, or invoke Lambda functions as part of its context. To use it, ensure your AWS credentials are available to the server (through environment variables or the configuration you provide in your editor) and start the MCP server so Zed can connect to localhost. In Zed, configure your MCP client to point at the local MCP server and the AI assistant will be able to leverage AWS context during conversations and code generation.

Prerequisites:

• Node.js 16+ (with npm) installed on your machine
• An AWS account with appropriate IAM permissions for the resources you want to access

Installation steps:

1. Install the MCP server package (example using npx):

   npx -y aws-mcp-server-for-zed

   This will install and start the MCP server locally, exposing the MCP endpoint for Zed.

2. Alternative local install (if you prefer):

   npm install -g aws-mcp-server-for-zed aws-mcp-server-for-zed

3. Configure AWS credentials for the server (one of the following):

• Environment variables (recommended for local development):

  export AWS_ACCESS_KEY_ID=<YOUR_AWS_ACCESS_KEY_ID> export AWS_SECRET_ACCESS_KEY=<YOUR_AWS_SECRET_ACCESS_KEY> export AWS_REGION=<YOUR_AWS_REGION>

• Or in Zed settings as a JSON block mapping the credentials to the MCP server (see integration docs).

4. Connect Zed to the local MCP server by pointing the MCP client to localhost (default MCP port). Verify the connection and start using AWS context in your assistant workflows.

Tips and common issues:

• Ensure AWS credentials have permissions to the resources you intend to access (S3, DynamoDB, Lambda, etc.).
• Keep credentials secure; prefer environment variables over hard-coding in settings.
• If you change AWS_REGION or credentials, restart the MCP server to pick up the new values.
• The MCP server runs locally; ensure your firewall allows local connections from Zed.
• Monitor for latency when invoking long-running AWS operations; consider caching results in your Zed workflows where appropriate.