We use essential cookies and similar tools that are necessary to provide our site and services. We use performance cookies to collect anonymous statistics, so we can understand how customers use our site and make improvements. Essential cookies cannot be deactivated, but you can choose “Customize†or “Decline†to decline performance cookies.
If you agree, AWS and approved third parties will also use cookies to provide useful site features, remember your preferences, and display relevant content, including relevant advertising. To accept or decline all non-essential cookies, choose “Accept†or “Decline.†To make more detailed choices, choose “Customize.â€Â
Customize cookie preferences
We use cookies and similar tools (collectively, "cookies") for the following purposes.
Essential
Essential cookies are necessary to provide our site and services and cannot be deactivated. They are usually set in response to your actions on the site, such as setting your privacy preferences, signing in, or filling in forms.
Performance
Performance cookies provide anonymous statistics about how customers navigate our site so we can improve site experience and performance. Approved third parties may perform analytics on our behalf, but they cannot use the data for their own purposes.
Allowed
Functional
Functional cookies help us provide useful site features, remember your preferences, and display relevant content. Approved third parties may set these cookies to provide certain site features. If you do not allow these cookies, then some or all of these services may not function properly.
Allowed
Advertising
Advertising cookies may be set through our site by us or our advertising partners and help us deliver relevant marketing content. If you do not allow these cookies, you will experience less relevant advertising.
Allowed
Blocking some types of cookies may impact your experience of our sites. You may review and change your choices at any time by selecting Cookie preferences in the footer of this site. We and selected third-parties use cookies or similar technologies as specified in the AWS Cookie Notice.
Your privacy choices
We display ads relevant to your interests on AWS sites and on other properties, including cross-context behavioral advertising. Cross-context behavioral advertising uses data from one site or app to advertise to you on a different company’s site or app.
To not allow AWS cross-context behavioral advertising based on cookies or similar technologies, select “Don't allow†and “Save privacy choices†below, or visit an AWS site with a legally-recognized decline signal enabled, such as the Global Privacy Control. If you delete your cookies or visit this site from a different browser or device, you will need to make your selection again. For more information about cookies and how we use them, please read our AWS Cookie Notice.
With the recent AWS endorsement of MCP (aka Model Context Protocol), many customers and partners have been trying to use the MCP protocol with Amazon Bedrock foundation models like Claude Sonnet 3.7 or Amazon Nova Pro. The "Run MCP via Bedrock Agents" blog shows how to implement this programmatically using Amazon Bedrock inline agents. If you want to explore an alternative method for setting up MCP-based agentic workflows using Amazon Bedrock with a GUI-based client (not Claude Desktop), this post will guide you through the setup process step by step. But first, let's try to understand some basics.
What is Model Context Protocol aka MCP?
MCP is an open protocol that enables seamless integration between AI apps & agents and your tools & data sources. Its like what API is for web applications i.e standardising the interaction between AI applications, tools and external data sources.
MCP allows AI to reach data instead of taking data to AI, reducing barriers for FMs to interact with data sources, tools and other applications allowing enterprise workflows that were not possible before
Official AWS MCP available servers : — follow setup guide here
Core MCP Server - Automatic MCP Server Management
Amazon Bedrock Knowledge Bases Retrieval MCP Server - A server for accessing Amazon Bedrock Knowledge Bases.
AWS CDK MCP Server - AWS CDK project analysis and assistance
Cost Analysis MCP Server - A server for AWS Cost Analysis.
Amazon Nova Canvas MCP Server - A server for generating images using Amazon Nova Canvas.
Why need a desktop client?
MCP desktop client allows a GUI based interface to setup MCP tools locally or 3P tools without needing to code. There are multiple desktop client's available most popular of those is Claude Desktop client. Only limitation with Claude desktop client is it does not allow you to choose your own bedrock models or interact with Amazon Bedrock. There are various open source clients available including Goose, Cursor, 5ire etc. and for this post we will be using Goose to connect Amazon Bedrock models. Here is an exhaustive list of other available MCP clients that you can use
Setup AWS CLI --- Region US-east-1 — add your profile (do use not goose config). You can follow instructions here to setup CLI.
Configure Amazon Bedrock as model provider , for this you will AWS_Profile that you can find here on how to set it up. (Do not use Goose Configure for setting up Bedrock models)
Image not found
Choose Amazon Bedrock as you model provider
Select model provider to Amazon Bedrock , add model using model ID : amazon.nova-pro-v1:0 — Try Nova, its cost effective us.anthropic.claude-3-7-sonnet-20250219-v1:0 — for Claude Sonnet 3.7 or choose any other model available via Amazon Bedrock
Awesome, you have a desktop client running with Amazon bedrock models you can toggle between models
Image not found
Toggle between Amazon Bedrock models
As we have basic desktop client setup using Amazon bedrock, now its time to setup MCP tools. Click "Add extension"
Image not found
Add MCP servers
Choose MCP servers based on the workflows/ usecase you want to test it for from here. Over 100+ MCP servers are available including official AWS MCP server repo. Try adding Slack , pgvector, sequential reasoning and much more. For this example let's add sequential thinking i.e giving deep thinking ability as a tool for your local client
Go to official model context protocol (mcp server repo) and choose filesystem MCP server : https://github.com/modelcontextprotocol/servers/tree/main/src/filesystem , follow readme.md and go to claude desktop setup instructions. something like given below.
Now this is a setup where setup is different from Claude desktop and we need just command , here is how you can extract the command for adding MCP on Goose
Now go to Settings > click add custom extension and ID : filesystem Name : filesystem Description : Model Context Protocol (MCP) for filesystem operations. and under command - use the command we just extracted above , click add -- voila you have your first MCP setup
Image not found
Add custom MCP server
Now, let's test it out. Open a new session , toggle between Nova Pro or Claude 3.7 sonnet to test our amazon bedrock models are connected and give a test prompt. Here is an example below :
Image not found
Test MCP tool
Now we have our MCP based Agent setup using Amazon Bedrock.
Next Steps
try to setup AWS MCP servers , AWS MCP available servers :—
Core MCP Server - Automatic MCP Server Management
Amazon Bedrock Knowledge Bases Retrieval MCP Server - A server for accessing Amazon Bedrock Knowledge Bases.
AWS CDK MCP Server - AWS CDK project analysis and assistance
Cost Analysis MCP Server - A server for AWS Cost Analysis.
Amazon Nova Canvas MCP Server - A server for generating images using Amazon Nova Canvas.
Try adding AWS cost analyser MCP , to have your own AWS cost anlyser agent or setup Amazon Nova Canvas MCP to generate images using Amazon Nova Canvas.
Amazon Bedrock Knowledge Bases Retrieval MCP Server - A server for accessing Amazon Bedrock Knowledge Bases.
AWS CDK MCP Server - AWS CDK project analysis and assistance
Cost Analysis MCP Server - A server for AWS Cost Analysis.
Amazon Nova Canvas MCP Server - A server for generating images using Amazon Nova Canvas.
Do let us know in comments how you find this post and what you tried out next.
Contributor(s) : Ninad Joshi is a GenAI Partner Solution Architect, EMEA helping AWS partners build and scale GenAI workloads. In his free time Ninad likes playing Squash, travel and read or just simply hack personal prototypes. Ninad is a GenAI specialist with background in AI/ML, having working with consulting, startup and enterprise customers across various industry verticals.
Any opinions in this post are those of the individual author and may not reflect the opinions of AWS.
