Agent Skills
› NeverSight/learn-skills.dev
› deeplake-memory
deeplake-memory
GitHubDeeplake持久化内存的SDK和CLI工具，提供云端文件系统支持。用于挂载、管理跨会话持久存储，同步多AI代理数据，并执行初始化、登录、挂载及组织管理等操作。
Trigger Scenarios
用户需要跨会话保存或读取文件
用户希望实现多个AI助手间的数据实时共享
用户需要初始化或配置Deeplake云存储连接
用户需要管理Deeplake的组织、工作区或成员权限
Install
npx skills add NeverSight/learn-skills.dev --skill deeplake-memory -g -y
SKILL.md
Frontmatter
{
"name": "deeplake-memory",
"description": "SDK and CLI for Deeplake persistent memory — a cloud-backed filesystem for AI agents. Use when users want to mount, manage, or use persistent memory that survives across sessions, or run any `deeplake` CLI command."
}
Deeplake Memory — Persistent Filesystem for AI Agents
Deeplake Memory is a FUSE-mounted folder backed by Deeplake Cloud. Everything written here survives across sessions and syncs in real-time to all connected AI agents (Claude Code, Cursor, Copilot, etc.).
Memory Location
Your persistent memory is mounted at ~/.deeplake/memory/.
- Read/write files with standard commands (
cat,ls,echo >) - Sessions persist — files are shared across all Claude Code sessions
- Multi-agent — other AI assistants see the same files in real-time
- Cloud-backed — data stored in Deeplake Cloud, accessible from any machine
CLI Reference
Getting Started
# First time setup (interactive)
deeplake init
# Login to Deeplake
deeplake login
# Mount all registered filesystems
deeplake mount --all
# Check what's running
deeplake list
Commands
| Command | Description |
|---|---|
deeplake init [options] |
Initialize a new mount (interactive setup) |
deeplake mount <path|--all> |
Mount a specific path or all registered mounts |
deeplake umount <path|--all> |
Unmount a specific path or all mounts |
deeplake list |
Show all registered mounts and their status |
deeplake login |
Authenticate with Deeplake |
deeplake logout |
Clear stored credentials |
deeplake organization list |
List all organizations |
deeplake organization set <name|id> |
Switch active organization |
deeplake workspace list |
List all workspaces |
deeplake workspace set <name|id> |
Switch active workspace |
deeplake member invite <email|user> |
Invite a user to your organization |
deeplake member remove <email|user> |
Remove a user from your organization |
deeplake import <source-dir> <table> |
Bulk import a directory into a table |
deeplake version |
Show CLI version |
deeplake help |
Show help message |
Aliases
| Alias | Expands to |
|---|---|
org |
organization |
ws |
workspace |
stop |
umount |
status |
list |
--version |
version |
-h, --help |
help |
Init Options
deeplake init --org my-org --workspace prod --table my_project --path /path/to/mount
| Flag | Description | Default |
|---|---|---|
--org <name|id> |
Organization name or ID | from credentials |
--workspace <name> |
Workspace name | default |
--table <name> |
Table name | memory |
--path <path> |
Mount path (skips interactive prompt) | — |
--hook |
Auto-install memory sync hook (skips prompt) | — |
--skill <local|global> |
Install Deeplake skill for Claude Code (skips prompt) | — |
--auto |
Use all defaults, skip all prompts (implies --hook --skill global) | — |
Member Invite Options
deeplake member invite user@example.com --role writer
| Flag | Description | Default |
|---|---|---|
--role <admin|writer|reader> |
Access level for the invited user | writer |
Global Options
| Flag | Description |
|---|---|
-v, --verbose |
Enable verbose logging |
Environment Variables
| Variable | Description | Default |
|---|---|---|
DEEPLAKE_API_URL |
Deeplake API URL | https://api.deeplake.ai |
DEEPLAKE_TOKEN |
API token (for managed service) | — |
DEEPLAKE_ORG_ID |
Organization ID | — |
DEEPLAKE_WORKSPACE_ID |
Workspace ID | — |
DEEPLAKE_MOUNT_PATH |
Default mount path | — |
Examples
# First time setup (interactive)
deeplake init
# Fully automatic setup — all defaults, no prompts
deeplake init --auto
# Custom org and workspace
deeplake init --org my-org --workspace my-ws --table my_project
# Non-interactive with custom path + hooks + skill
deeplake init --org my-org --path ~/my-memory --hook --skill global
# Install skill only for this project
deeplake init --hook --skill local
# Switch organizations (re-points all memory mounts)
deeplake organization list
deeplake org set my-org
# Switch workspaces
deeplake workspace list
deeplake ws set production
# Invite a teammate
deeplake member invite user@example.com --role writer
# Revoke access
deeplake member remove user@example.com
# Unmount everything
deeplake umount --all
# Bulk import a directory (~400x faster than cp)
deeplake import ./my-data my_table
Version History
- e0220ca Current 2026-07-05 23:03
