{"id":47566457,"url":"https://github.com/aws-samples/sample-strands-agent-with-agentcore","last_synced_at":"2026-04-14T01:00:35.239Z","repository":{"id":311717399,"uuid":"1037029275","full_name":"aws-samples/sample-strands-agent-with-agentcore","owner":"aws-samples","description":"Reference architecture for agentic AI chatbots with Strands Agents and Amazon Bedrock AgentCore","archived":false,"fork":false,"pushed_at":"2026-04-13T21:15:07.000Z","size":62699,"stargazers_count":134,"open_issues_count":3,"forks_count":44,"subscribers_count":8,"default_branch":"main","last_synced_at":"2026-04-13T23:24:16.067Z","etag":null,"topics":["a2a-protocol","agent-to-agent","agentcore","ai-agents","amazon-bedrock","browser-automation","mcp","multi-agent","skills","strands-agents","voice-assistant"],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/aws-samples.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":"CODE_OF_CONDUCT.md","threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-08-13T00:44:01.000Z","updated_at":"2026-04-13T21:15:05.000Z","dependencies_parsed_at":"2025-08-26T07:11:47.869Z","dependency_job_id":"b02e4010-7f0a-4958-b8cb-5db9d5bc84e1","html_url":"https://github.com/aws-samples/sample-strands-agent-with-agentcore","commit_stats":null,"previous_names":["aws-samples/sample-strands-agent-chatbot","aws-samples/sample-strands-agent-with-agentcore"],"tags_count":1,"template":false,"template_full_name":"amazon-archives/__template_MIT-0","purl":"pkg:github/aws-samples/sample-strands-agent-with-agentcore","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aws-samples%2Fsample-strands-agent-with-agentcore","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aws-samples%2Fsample-strands-agent-with-agentcore/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aws-samples%2Fsample-strands-agent-with-agentcore/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aws-samples%2Fsample-strands-agent-with-agentcore/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/aws-samples","download_url":"https://codeload.github.com/aws-samples/sample-strands-agent-with-agentcore/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aws-samples%2Fsample-strands-agent-with-agentcore/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31777348,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-14T00:11:49.126Z","status":"ssl_error","status_checked_at":"2026-04-14T00:10:29.837Z","response_time":93,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["a2a-protocol","agent-to-agent","agentcore","ai-agents","amazon-bedrock","browser-automation","mcp","multi-agent","skills","strands-agents","voice-assistant"],"created_at":"2026-03-30T06:00:24.600Z","updated_at":"2026-04-14T01:00:35.228Z","avatar_url":"https://github.com/aws-samples.png","language":"Python","funding_links":[],"categories":["Community Projects"],"sub_categories":["For PyPI Packages"],"readme":"# Strands Agent Chatbot with Amazon Bedrock AgentCore\n\nAn end-to-end reference architecture for building **agentic workflows** using  \n**Strands Agents** and **Amazon Bedrock AgentCore**.\n\nThis repository demonstrates how to design and deploy a multi-agent chatbot that combines\ntool execution, memory, browser automation, and agent-to-agent collaboration.\nIt is intended as a realistic, extensible sample for teams exploring advanced agent architectures on AWS.\n\n---\n\n## Demo\n\nhttps://github.com/user-attachments/assets/11b383c2-2e14-4135-833f-b0b2bce62953\n\n\u003e **[Full demo (6 min)](https://drive.google.com/file/d/1Hk4hgiqqMntNRUN1xDkwcbJUA8MVEJEA/view?usp=sharing)**\n\n### Mobile App\n\n\u003ctable\u003e\n  \u003ctr\u003e\n    \u003ctd align=\"center\"\u003e\u003cimg src=\"docs/images/claude-code-agent.png\" width=\"180\"\u003e\u003cbr\u003e\u003csub\u003eCoding Agent\u003c/sub\u003e\u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\u003cimg src=\"docs/images/finance-analysis.png\" width=\"180\"\u003e\u003cbr\u003e\u003csub\u003eFinancial Analysis\u003c/sub\u003e\u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\u003cimg src=\"docs/images/browser-automation.png\" width=\"180\"\u003e\u003cbr\u003e\u003csub\u003eBrowser Automation\u003c/sub\u003e\u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\u003cimg src=\"docs/images/browser-automation-2.png\" width=\"180\"\u003e\u003cbr\u003e\u003csub\u003eWeb Search\u003c/sub\u003e\u003c/td\u003e\n  \u003c/tr\u003e\n\u003c/table\u003e\n\n---\n\n## What this repository demonstrates\n\n- Multi-agent orchestration with **Strands Agents**\n- Integration with **Amazon Bedrock AgentCore**\n- Tool-enabled agents (search, finance, weather, browser, code interpreter)\n- Autonomous browser, documentation, and analysis workflows\n- Modular architecture adaptable to real customer use cases\n- Infrastructure-as-Code (CDK) for repeatable deployment\n\nIf you are building **agentic AI applications on AWS** and want a concrete, end-to-end example,\nthis repository is designed to be read, run, and extended.\n\n**Quick links:**  \n[Architecture](#architecture-overview) · [Key Features](#key-features) · [Quick Start](#quick-start)\n\n---\n\n## Architecture Overview\n\nThis sample combines **Strands Agent orchestration** with **Amazon Bedrock AgentCore services**:\n\n| Component | Role |\n|-----------|------|\n| **Strands Agents** | Multi-turn reasoning and tool orchestration |\n| **AgentCore Runtime** | Managed, containerized agent execution |\n| **AgentCore Memory** | Persistent conversation state and summarization |\n| **AgentCore Gateway** | MCP-based tool integration with SigV4 authentication |\n| **AgentCore Code Interpreter** | Secure Python execution for analysis and document generation |\n| **AgentCore Browser** | Headless browser automation with live view |\n| **AgentCore Identity** | End-user authentication and 3LO OAuth delegation |\n| **AgentCore Observability** | Trace collection and agent execution monitoring |\n| **Amazon Nova Act** | Visual reasoning model for browser automation |\n\n\u003cimg src=\"docs/images/architecture-overview.png\"\n     alt=\"Architecture Overview\"\n     width=\"1200\"\u003e\n\n---\n\n## Key Features\n\n- Strands-based agent orchestration\n- Amazon Bedrock AgentCore Runtime, Gateway, and Memory\n- MCP Gateway tools (Google, Wikipedia, ArXiv, Yahoo Finance, Tavily, Open-Meteo)\n- Agent-to-Agent (A2A) collaboration — including remote **Claude Agent SDK (Claude Code)** for agentic coding tasks\n- Built-in Code Interpreter for charts and documents\n- Multimodal input and output (vision, charts, documents, screenshots)\n- Real-time voice interaction with Amazon Nova Sonic 2\n\n---\n\n## Skill System (Progressive Disclosure)\n\nTools are organized into **skills** — grouped units with SKILL.md instructions that the agent loads on demand.\nInstead of injecting all tool documentation into every prompt, the agent activates only the skills it needs:\n\n1. **L1 Catalog** — Skill names and one-line descriptions (always in system prompt)\n2. **L2 Instructions** — Full SKILL.md loaded via `skill_dispatcher` when activated\n3. **L3 Execution** — Tool calls via `skill_executor`\n\nThis keeps prompt size small while giving the agent access to detailed instructions when needed.\n\nDesign notes:\n- https://medium.com/towards-artificial-intelligence/agent-skills-part-2-bridging-skills-with-production-tool-ecosystems-422e4a63fcad\n\n```\nskills/\n├── visual-design/          # Charts, posters, infographics (Code Interpreter)\n├── code-interpreter/       # General code execution\n├── browser-automation/     # Nova Act browser tools\n├── word-documents/         # Word document generation\n├── excel-spreadsheets/     # Excel spreadsheet generation\n├── powerpoint-presentations/  # PowerPoint generation\n├── gmail/                  # Gmail read/search/delete (3LO OAuth)\n├── google-calendar/        # Calendar events (3LO OAuth)\n├── notion/                 # Notion pages and databases (3LO OAuth)\n├── github/                 # GitHub repos, issues, PRs, code (3LO OAuth)\n└── ...                     # 20+ skills (web search, finance, maps, weather, and more)\n```\n\n---\n\n## Multi-Protocol Tool Architecture\n\n| Tool Category | Protocol | Examples | Authentication |\n|--------------|----------|----------|----------------|\n| Local Tools | Direct Python | Web Search, URL Fetcher, Visualization | None |\n| Built-in Tools | AWS SDK / WebSocket | Code Interpreter, Browser (Nova Act) | IAM |\n| Gateway Tools | MCP | Google Search, Maps, Wikipedia, ArXiv, Finance | SigV4 |\n| Private API Tools | MCP (3LO OAuth) | Gmail, Google Calendar, Notion, GitHub | OAuth 2.0 |\n| A2A Tools | A2A | Research Agent, Browser-Use Agent | SigV4 |\n\nTotal: **100+ tools across 20 tool groups**\nSee [docs/guides/TOOLS.md](docs/guides/TOOLS.md) for full details.\n\n---\n\n## Voice Mode\n\nReal-time voice interaction using **Amazon Nova Sonic 2**, **Strands BidiAgent**,\nand **AgentCore Runtime WebSocket**.\n\n- Seamless switching between voice and text within a single session\n- Shared conversation history across both modes\n- Full tool execution support during voice conversations\n\n---\n\n## Memory Architecture and Long-Context Management\n\nThe system uses **AgentCore Memory** with:\n- Short-term session memory\n- Long-term summarized memory\n\nLong conversations are compacted using a context summarization strategy to\nretain key information while controlling token growth.\n\nDesign notes:\n- https://medium.com/@revoir07/long-context-compaction-for-ai-agents-part-1-design-principles-2bf4a5748154\n\n---\n\n## Token Optimization via Prompt Caching\n\nPrompt caching reduces input token usage by reusing system prompts, stable instruction blocks,\nand repeated conversation context across agent loop iterations.\n\nThis project originally implemented caching via custom Strands hooks.\nThe approach has since been upstreamed into the Strands SDK as a built-in feature\n([strands-agents/sdk-python#1438](https://github.com/strands-agents/sdk-python/pull/1438)):\n\n```python\nfrom strands.models import BedrockModel, CacheConfig\n\nmodel = BedrockModel(\n    model_id=\"us.anthropic.claude-sonnet-4-6-v1\",\n    cache_config=CacheConfig(strategy=\"auto\")\n)\n```\n\nDesign notes:\n- https://medium.com/@revoir07/agent-loop-caching-the-missing-optimization-for-agent-workflows-230cc530eb72\n\n---\n\n## Multi-Agent Architecture\n\nAgent-to-Agent communication is handled via the **A2A protocol**, allowing the supervisor agent to delegate tasks to specialized worker agents such as a deep research agent.\nMultiple agents collaborating in sequence — each handling its own role — are coordinated using the **Swarm** pattern.\n\nThis architecture also includes a **Claude Agent SDK (Claude Code)** instance deployed as a remote A2A agent on **AgentCore Runtime**.\nThe supervisor delegates agentic coding tasks — multi-file implementation, refactoring, test suites — to this agent over A2A, with an S3-backed workspace that persists files across sessions.\n\nDesign notes:\n- https://medium.com/@revoir07/extend-your-chatbot-with-deep-research-using-a2a-ba4de3ed23e9\n\n---\n\n## Use Cases\n\n- Financial research agents\n- Technical research assistants using multi-agent patterns\n- Autonomous web automation agents\n- Memory-backed conversational assistants\n- Hybrid research workflows using MCP, A2A, and AWS SDK tools\n\n---\n\n## Quick Start\n\n### Prerequisites\n\n- AWS account with Bedrock access  \n- AWS CLI configured  \n- Docker installed  \n- Node.js 18+ and Python 3.13+  \n\n---\n\n### Local Development\n\n```bash\ngit clone https://github.com/aws-samples/sample-strands-agent-with-agentcore.git\ncd sample-strands-agent-with-agentcore\n\ncd chatbot-app\n./setup.sh\n\ncd ../agent-blueprint\ncp .env.example .env\n# configure credentials\n\ncd ../chatbot-app\n./start.sh\n```\n\nFrontend will be available at http://localhost:3000.\n\n---\n\n### Cloud Deployment\n\nThe interactive deployment script supports:\n\t1.\tAgentCore Runtime\n\t2.\tFrontend + BFF\n\t3.\tMCP Gateway tools\n\t4.\tA2A runtimes\n\t5.\tFull stack deployment\n```\ncd agent-blueprint\n./deploy.sh\n```\n\n### Project Structure\n```\nsample-strands-agent-chatbot/\n├── chatbot-app/\n│   ├── frontend/\n│   └── agentcore/\n└── agent-blueprint/\n    ├── chatbot-deployment/\n    ├── agentcore-gateway-stack/\n    ├── agentcore-runtime-stack/\n    └── agentcore-runtime-a2a-stack/\n```\n\n## Documentation\n\n- DEPLOYMENT.md\n- docs/guides/TROUBLESHOOTING.md\n\n---\n\n## License\n\nMIT License. See LICENSE for details.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faws-samples%2Fsample-strands-agent-with-agentcore","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Faws-samples%2Fsample-strands-agent-with-agentcore","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faws-samples%2Fsample-strands-agent-with-agentcore/lists"}