AWS location documentation change
Summary
Expanded documentation for configuring LLM clients to use MCP servers and context files. Added detailed configuration examples for multiple tools (Kiro, VSCode Copilot, Cline, Cursor, Claude Code, Gemini) including server URLs, command parameters, and environment variables. Enhanced explanations about context optimization to reduce hallucinations and improve accuracy.
Security assessment
The change adds security documentation by explicitly recommending READ_OPERATIONS_ONLY=true in AWS API MCP configurations, which limits operations to read-only access. This reduces attack surface by preventing unintended modifications. However, there's no evidence this addresses a specific security incident.
Diff
diff --git a/location/latest/developerguide/ai-llms.md b/location/latest/developerguide/ai-llms.md index cf4a0625d..1df90049d 100644 --- a//location/latest/developerguide/ai-llms.md +++ b//location/latest/developerguide/ai-llms.md @@ -9 +9 @@ Recommended MCP ServersUseful ContextKiro Agent Configuration -AI and LLMs can significantly accelerate development with Amazon Location Service by providing intelligent assistance for API usage, code generation, and troubleshooting. By configuring your LLM client with the right MCP servers and context, you can create a powerful development assistant that understands AWS services and Amazon Location Service specifics. +AI and LLMs can significantly accelerate development with Amazon Location Service by providing intelligent assistance for API usage, code generation, and troubleshooting. By configuring your LLM client with the right MCP servers and context, you can create a powerful development assistant that understands AWS services and Amazon Location Service specifics. Using a minimal context and MCP configuration as recommended on this page can ensure your LLM model of choice has enough context to lead to correct results without overwhelming the context window. This can reduce hallucinations and increase result accuracy. This configuration also ensures that model knowledge cutoff does not impact the quality of the results. @@ -15 +15 @@ Model Context Protocol (MCP) servers extend LLM capabilities by providing access - * **aws-knowledge-mcp-server** \- Access to AWS documentation, API references, best practices, and knowledge bases. Does not require AWS credentials or authentication, making it ideal for documentation lookup without credential management. [Learn more](https://awslabs.github.io/mcp/servers/aws-knowledge-mcp-server) + * **[aws-knowledge-mcp-server](https://awslabs.github.io/mcp/servers/aws-knowledge-mcp-server)** \- Access to AWS documentation, API references, best practices, and knowledge bases. Does not require AWS credentials or authentication, making it ideal for documentation lookup without credential management. @@ -17 +17 @@ Model Context Protocol (MCP) servers extend LLM capabilities by providing access - * **aws-api-mcp-server** \- Direct AWS API interactions and CLI command execution. Requires AWS credentials. [Learn more](https://awslabs.github.io/mcp/servers/aws-api-mcp-server) + * **[aws-api-mcp-server](https://awslabs.github.io/mcp/servers/aws-api-mcp-server)** \- Direct AWS API interactions and CLI command execution. Requires AWS credentials. @@ -26 +26,4 @@ Configure your LLM client with the MCP servers using the appropriate configurati -#### One-click install +Kiro + + +**One-click install:** @@ -35 +38 @@ Configure your LLM client with the MCP servers using the appropriate configurati -#### Manual configuration +**Manual configuration:** @@ -37 +40 @@ Configure your LLM client with the MCP servers using the appropriate configurati -To add these MCP servers to your Kiro agent configuration, use the following format: +Add the following to your Kiro agent configuration. For more information about [Kiro configuration](https://kiro.dev/docs/mcp/configuration/), see the Kiro documentation. @@ -56,0 +60,122 @@ To add these MCP servers to your Kiro agent configuration, use the following for +VSCode with Copilot + + +**One-click install:** + + * [AWS Knowledge MCP Server](https://vscode.dev/redirect/mcp/install?name=aws-knowledge-mcp&config=%7B%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fknowledge-mcp.global.api.aws%22%7D) + + * [AWS API MCP Server](https://insiders.vscode.dev/redirect/mcp/install?name=AWS%20API%20MCP%20Server&config=%7B%22command%22%3A%22uvx%22%2C%22args%22%3A%5B%22awslabs.aws-api-mcp-server%40latest%22%5D%2C%22env%22%3A%7B%22AWS_REGION%22%3A%22us-east-1%22%7D%2C%22type%22%3A%22stdio%22%7D) + + + + +**Manual configuration:** + +Add the following to your VSCode mcp.json file. For more information about [MCP servers in VS Code](https://code.visualstudio.com/docs/copilot/customization/mcp-servers), see the VSCode documentation. + + + { + "servers": { + "aws-knowledge-mcp-server": { + "type": "http", + "url": "https://knowledge-mcp.global.api.aws" + }, + "aws-api-mcp-server": { + "command": "uvx", + "args": ["awslabs.aws-api-mcp-server@latest"], + "env": { + "AWS_REGION": "us-east-1", + "READ_OPERATIONS_ONLY": "true" + } + } + } + } + +VSCode with Cline + + +**Manual configuration:** + +Add the following to your Cline MCP settings file (cline_mcp_settings.json). For more information about [Cline MCP configuration](https://docs.cline.bot/mcp/configuring-mcp-servers), see the Cline documentation. + + + { + "mcpServers": { + "aws-knowledge-mcp-server": { + "type": "streamableHttp", + "url": "https://knowledge-mcp.global.api.aws" + }, + "aws-api-mcp-server": { + "type": "stdio", + "command": "uvx", + "args": ["awslabs.aws-api-mcp-server@latest"], + "env": { + "AWS_REGION": "us-east-1", + "READ_OPERATIONS_ONLY": "true" + } + } + } + } + +Cursor + + +**Manual configuration:** + +Add the following to your Cursor mcp.json file. For more information about [Cursor MCP configuration](https://cursor.com/docs/context/mcp), see the Cursor documentation. + + + { + "mcpServers": { + "aws-knowledge-mcp-server": { + "url": "https://knowledge-mcp.global.api.aws" + }, + "aws-api-mcp-server": { + "command": "uvx", + "args": ["awslabs.aws-api-mcp-server@latest"], + "env": { + "AWS_REGION": "us-east-1", + "READ_OPERATIONS_ONLY": "true" + } + } + } + } + +Claude Code + + +**Manual configuration:** + +Add MCP servers using the Claude CLI commands. For more information about [Claude Code MCP setup](https://code.claude.com/docs/en/mcp), see the Claude Code documentation. + + + # Add AWS Knowledge MCP Server (HTTP) + claude mcp add --transport http aws-knowledge-mcp-server https://knowledge-mcp.global.api.aws + + # Add AWS API MCP Server (stdio) + claude mcp add --transport stdio aws-api-mcp-server -- uvx awslabs.aws-api-mcp-server@latest + +Gemini Code Assist + + +**Manual configuration:** + +Add the following to your Gemini settings JSON file (~/.gemini/settings.json). For more information about [Gemini Code Assist MCP configuration](https://developers.google.com/gemini-code-assist/docs/use-agentic-chat-pair-programmer#configure-mcp-servers), see the Google Cloud documentation. + + + { + "mcpServers": { + "aws-knowledge-mcp-server": { + "httpUrl": "https://knowledge-mcp.global.api.aws" + }, + "aws-api-mcp-server": { + "command": "uvx", + "args": ["awslabs.aws-api-mcp-server@latest"], + "env": { + "AWS_REGION": "us-east-1", + "READ_OPERATIONS_ONLY": "true" + } + } + } + } + @@ -63 +188 @@ There is a maintained [AGENTS.md](https://github.com/aws-geospatial/amazon-locat -To add this context to your Kiro agent configuration: +To use this context file, first download it locally: @@ -65 +189,0 @@ To add this context to your Kiro agent configuration: - 1. Download the AGENTS.md file locally to your desired context path: @@ -69 +193,7 @@ To add this context to your Kiro agent configuration: - 2. Add the local file to your agent configuration: +Then configure your LLM client to use the downloaded file: + +Kiro + + +Add the local file to your agent configuration: + @@ -76,0 +207,21 @@ To add this context to your Kiro agent configuration: +VSCode with Copilot + + +Place the downloaded AGENTS.md file at the root of your workspace. VSCode will automatically apply the instructions to all chat requests. To enable this feature, ensure the [chat.useAgentsMdFile](vscode://settings/chat.useAgentsMdFile) setting is enabled. For more information, see [custom instructions](https://code.visualstudio.com/docs/copilot/customization/custom-instructions) in the VSCode documentation. + +VSCode with Cline + + +Place the downloaded AGENTS.md file in your project root or use @ mentions to reference it in your conversations. Cline will automatically discover project files and you can reference the context using `@AGENTS.md` in your prompts. For more information about [context management](https://docs.cline.bot/prompting/understanding-context-management), see the Cline documentation. + +Cursor + + +Use @ mentions to reference the downloaded AGENTS.md file in your conversations. You can reference files using `@Files & Folders` and then search for the AGENTS.md file, or drag the file directly into the chat. For more information about [@ mentions](https://cursor.com/docs/context/mentions), see the Cursor documentation. + +Claude Code + + +Add the downloaded AGENTS.md file to your project directory. You can include it in your project's CLAUDE.md file or reference it directly in your current session. For more information about [Claude Code MCP setup](https://code.claude.com/docs/en/mcp), see the Claude Code documentation. + +Gemini Code Assist @@ -78,0 +230 @@ To add this context to your Kiro agent configuration: +Create a GEMINI.md file in your project root or ~/.gemini/GEMINI.md for global context, and include the contents of the downloaded AGENTS.md file. For more information about [context files](https://developers.google.com/gemini-code-assist/docs/use-agentic-chat-pair-programmer#create-a-context-file), see the Google Cloud documentation.