Managed Deep Agents - Instructions and Context Hub @LangChain
Managed Deep Agents - Instructions and Context Hub  @LangChain
Uploaded August 2026 | Updated November 2026, 3 days ago
Instructions define the agent’s behavior. They make up the core of the agent’s system prompt. You can define them in a simple markdown file and they are picked up automatically by the agent.

When you run mda deploy to deploy the agent, instructions are automatically synced to the agent’s Context Hub repo. You can then edit the instructions in the LangSmith UI and have your changes automatically propagated to the agent.

Docs: docs.langchain.com/langsmith/python/managed-deep-agents-instructions
LangSmith: smith.langchain.com
Managed Deep Agents - Instructions and Context HubDeep Agents by LangChain explained in 1 minutePodiums Playbook for Scaling AI Agents Org-WideModel Fallback Middleware (Python)Interrupt 26: The Agent Conference by LangChainInterrupt NYC: Opening KeynoteObserving And Testing CX Agents | Interrupt 26The Self-Driving Eval Trick No AI Benchmark BeatsThe Etsy Gifting Assistant: From Prototype to Production | Interrupt 26What building Teslas autopilot chips taught Cognitions President about AIHow Madrigal Pharmaceuticals Cut Time to Production From 12 Weeks to 2 with LangChain & LangSmithWhat is Deep Agents? An open source harness for long-running agents.
LangChain |

Managed Deep Agents - Instructions and Context Hub

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER