deep-agents-memory

Manage AI agent memory and filesystem interactions through pluggable storage backends.

3|Updated Jan 20, 2026

Search Tags:#state management #backend #persistence #filesystem #agent memory #deep agents

Ready to run

npx skills add https://github.com/Sameeh07/AGENT-SKILLS --skill deep-agents-memory-sameeh07

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: deep-agents-memory
Source: https://github.com/Sameeh07/AGENT-SKILLS/tree/main/skills/deep-agents-memory
Command: npx skills add https://github.com/Sameeh07/AGENT-SKILLS --skill deep-agents-memory-sameeh07

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill addresses the need for AI agents to manage persistent memory, temporary files, and interact with the filesystem, enabling more complex and stateful operations.

Core Features & Use Cases

StateBackend: Handles ephemeral, thread-local file storage for short-term data.
StoreBackend: Provides persistent storage across threads and sessions for long-term memory.
FilesystemMiddleware: Offers tools like ls, read_file, write_file, edit_file, glob, and grep for direct filesystem interaction.
CompositeBackend: Allows routing different file paths to different storage backends (ephemeral, persistent, or filesystem).
Use Case: An agent needs to save user preferences persistently while using temporary files for intermediate calculations within a single task.

Quick Start

Configure the agent to use a CompositeBackend that routes files under /memories/ to a persistent StoreBackend and all other files to the default ephemeral StateBackend.

Frequently Asked Questions about deep-agents-memory

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I persist agent memory across multiple sessions?▼

Agent memory persistence across sessions is handled by a StoreBackend, which provides long-term storage for cross-session data. It allows agents to save user preferences and stateful information permanently.

What is the best way to manage temporary files for intermediate agent calculations?▼

Temporary files for intermediate calculations are managed using a StateBackend, which handles ephemeral, thread-local file storage. This provides short-term data access without permanent disk footprint.

Can I route different file paths to different storage backends for my AI agent?▼

Routing different file paths to different storage backends is achieved using a CompositeBackend. It directs files under specific paths like `/memories/` to persistent storage while sending others to ephemeral storage.

What filesystem tools are available for agent state management and file access?▼

Filesystem access for agent state management includes tools like `ls`, `read_file`, `write_file`, `edit_file`, `glob`, and `grep`. These are provided by the FilesystemMiddleware for direct disk interaction.

How do I configure an agent to use both persistent memory and temporary file storage?▼

Configuring an agent for both persistent and temporary storage involves setting up a CompositeBackend. This routes specific memory paths to a persistent StoreBackend while defaulting other files to an ephemeral StateBackend.