Bedrock Chat with MCP

Example Use Cases

• 1

  Call external tools or services (via MCP) to augment LLM responses with live fetches, transformations, or other tool outputs.

• 2

  A configurable chat frontend with model selection, prompt caching, and persistent YAML chat history to resume or audit conversations.

• 3

  Integrate AWS Bedrock (or other LLMs) with tool-based context pipes using a local MCP server configuration.

Description

This repository implements a chat UI (Streamlit) that connects LLMs to MCP (Model Context Protocol) tools via a configurable MCP server description (config/mcp_config.json). It lets you select models, enable prompt caching, choose tools, upload files, and stores chat history in YAML for resumable conversations. The setup standardizes providing external context and tool access to LLMs, making it easier to augment model responses with tool outputs and custom workflows.