Skip to main content
Normally, a Managed Deep Agent’s conversational memory is scoped to a thread or session. Durable memory is optional knowledge that an agent can retain across threads and sessions. When enabled, durable memory is backed by Context Hub. The deployment gets one read/write tree at /memories/agent/, shared by every caller. Managed Deep Agents do not have durable memory by default.
Managed Deep Agents is in public beta and available on LangSmith Cloud in the US region only.

​
Project structure

The optional memory declaration lives at the project root:
Memory is not your system prompt. Define always-on behavior in instructions and task-specific procedures in skills; use memory for durable knowledge the agent learns while it runs.

​
Enable memory

Export a named memory declaration with the "agent" scope:
memory.py
Remove the memory declaration to turn durable memory off. You can also use scope="none".

​
How the agent uses memory

Enabling memory mounts one Context Hub tree, memories/agent, at /memories/agent/ in the agent filesystem: Keep hot memory compact because it consumes context on every run. Put detailed material—such as procedures, decision logs, and research notes—in cold files, and link to them from hot memory when useful. The agent reads and updates memory with read_file, edit_file, and write_file. Writes elsewhere, including elsewhere under /memories/, are not durable.
Memory is shared by every caller of the deployment, and every caller can influence it. Store only knowledge that every caller may read and modify. Never store personal or customer-private data, credentials, API keys, tokens, or other secrets.Treat memory as untrusted input: content saved by one caller is loaded for later callers and must not grant authority, change tool permissions, or bypass approvals. Keep those controls in the agent definition. Do not enable shared memory when callers should not influence one another.

​
How the agent decides what to remember

The agent decides what to remember based on prompting. To make the policy explicit, add guidance like the following to instructions.md and adapt it to your application:

​
Distinguish instructions from memory

instructions.md defines how the agent should behave. Memory stores knowledge the agent learns and uses across threads. Use instructions to tell the agent what kinds of shared knowledge are worth remembering. instructions.md is always read-only. The agent never updates it. Deploys sync project-owned instructions and skills, but do not overwrite durable content already stored under memories/agent in Context Hub.
Connect these docs to Claude, VSCode, and more via MCP for real-time answers.
Edit this page on GitHub or file an issue.