aws-samples / bedrock-claude-chatbot

What it does

• Agentic orchestration. A Strands Agent owns every conversation. Tools are opt-in per session from the sidebar; the orchestrator decides when to invoke them and relays sub-agent results. Live activity is streamed to the chat bubble (tool spi
• Broad model catalog, three engines. Model routing is registry-driven (model_id.json). Supports Bedrock Runtime models via the Converse API (Anthropic Claude, DeepSeek, Kimi, ...) and Bedrock Mantle OpenAI-compatible models via the Responses
• Advanced Data Analytics (sidebar tool). a data-analysis sub-agent writes and runs Python against your attached datasets in a sandbox, iterating on errors, validating its own charts (vision feedback), and returning a written analysis with in
• Document Generator & Editor (sidebar tool). a sub-agent produces and edits Word/PowerPoint/PDF/Excel files in the Code Interpreter sandbox (python-docx, python-pptx, reportlab, xlsxwriter, ...), verifies them structurally, and delivers them
• Research Workers (sidebar tool). the orchestrator fans self-contained research briefs out to parallel ephemeral worker agents (each with web search + page fetch), keeping its own context clean. Progress is shown per worker; citations are ag