# LangGraph on Amazon Bedrock AgentCore Runtime — companion artifacts

Companion artifacts for the FactualMinds post
**Building Production AI Agents with LangGraph and AWS Bedrock AgentCore**.

**Public URLs:** `https://www.factualminds.com/examples/architecture-blog-2026/langgraph-agentcore/{path}`

## Demo disclaimer

These files are **architecture diagrams and syntax-checkable stubs**, not a production deploy and **not** an anonymized client engagement.

- Account IDs, Memory IDs, model IDs, and Gateway ARNs are placeholders.
- Do **not** treat stub latencies or cost numbers as measured outcomes.
- First-party platform signals cited in the post (Gateway ~180→95 ms canary, AgentCore TCO worksheet) come from earlier published FactualMinds AgentCore posts — not from these stubs.

## Status snapshot (verify before you build)

Checked **August 27, 2026** against AWS and LangGraph docs:

| Capability | Status |
| ---------- | ------ |
| AgentCore Runtime hosts LangGraph via `BedrockAgentCoreApp` | Documented — [Use any agent framework](https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/using-any-agent-framework.html) |
| Official sample | [awslabs agentcore-samples langgraph](https://github.com/awslabs/amazon-bedrock-agentcore-samples/tree/main/03-integrations/agentic-frameworks/langgraph) |
| `AgentCoreMemorySaver` / `AgentCoreMemoryStore` | Documented — [Memory + LangGraph](https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/memory-integrate-lang.html) |
| Bedrock Converse via `init_chat_model(..., model_provider="bedrock_converse")` | Documented in the AWS LangGraph sample |
| AgentCore Harness as a LangGraph host | **Not** the path — Harness is config-driven; LangGraph runs on **Runtime** |
| Runtime architecture | **ARM64** containers; x86 images will not start |

## Versions (assumed)

Pin and verify in your account before deploy:

| Component | Assumed version / note |
| --------- | ---------------------- |
| Python | 3.12+ |
| `langgraph` | current Graph API (`StateGraph`, `START`, `END`, `add_conditional_edges`) |
| `langchain` | `init_chat_model` |
| `langchain-aws` | Bedrock Converse chat models |
| `langgraph-checkpoint-aws` | `AgentCoreMemorySaver` |
| `bedrock-agentcore` | `BedrockAgentCoreApp` |
| `boto3` | ≥ 1.38.0 |
| Default model in stub | Cross-region inference profile `us.anthropic.claude-sonnet-4-6` — **swap** to a model your account allows |

Local stubs compile without AWS. A deployed Runtime needs an IAM execution role, an ARM64 image or code zip, a supported region, and a Runtime endpoint.

## Layout

| Path | Purpose |
| ---- | ------- |
| [`architecture.drawio`](architecture.drawio) | Reference architecture: app → LangGraph Runtime → AgentCore → Bedrock → enterprise APIs, with HITL |
| [`langgraph_runtime_stub.py`](langgraph_runtime_stub.py) | `StateGraph` + `BedrockAgentCoreApp` entrypoint sketch |
| [`monday-checklist.md`](monday-checklist.md) | Ship-this-week gate list |

## Local checks

```bash
python3 -m py_compile langgraph_runtime_stub.py
```

## Related posts

- [LangGraph + AgentCore](https://www.factualminds.com/blog/langgraph-aws-bedrock-agentcore-2026/)
- [LangChain + Bedrock](https://www.factualminds.com/blog/langchain-aws-bedrock-2026/)
- [AgentCore production guide](https://www.factualminds.com/blog/amazon-bedrock-agentcore-production/)
- [AgentCore pricing calculator](https://www.factualminds.com/tools/amazon-bedrock-agentcore-pricing-calculator/)