AgentMemory CLI

Command-line interface for AgentMemory - persistent cloud memory for AI agents.

Installation

npm (recommended)

npm install -g agentmemory-cli

npx (no install)

npx agentmemory-cli store "Your memory"

curl installer

curl -fsSL https://agentmemory.cloud/install.sh | bash

Quick Start

1. Initialize with your API key:

agentmemory init
# Enter your API key from https://agentmemory.cloud/dashboard

2. Store a memory:

agentmemory store "User prefers dark mode and likes updates at 9 AM"

3. Search memories:

agentmemory search "user preferences"

Commands

| Command | Description | |---------|-------------| | agentmemory init | Configure your API key | | agentmemory store <content> | Store a new memory | | agentmemory search <query> | Semantic search | | agentmemory list | List all memories | | agentmemory delete <id> | Delete a memory | | agentmemory sync | Sync with local MEMORY.md | | agentmemory export | Export as JSON | | agentmemory import <file> | Import from JSON |

Options

Global Options

• --json - Output as JSON (for scripting)
• --help - Show help

Store Options

agentmemory store "content" --category preferences
agentmemory store "content" --metadata '{"priority": "high"}'

Search Options

agentmemory search "query" --limit 20

Sync Options

agentmemory sync              # Bidirectional sync
agentmemory sync --push       # Push local to cloud only
agentmemory sync --pull       # Pull cloud to local only
agentmemory sync --status     # Check sync status

Configuration

Configuration is stored in ~/.agentmemory/config.json:

{
  "api_key": "am_xxxxx",
  "api_url": "https://agentmemory.cloud/api",
  "memory_file": "./MEMORY.md"
}

Environment Variables

You can also use environment variables:

export AGENTMEMORY_API_KEY=am_xxxxx
export AGENTMEMORY_API_URL=https://agentmemory.cloud/api

MEMORY.md Sync

The CLI can sync with a local MEMORY.md file, enabling:

• Offline editing - Edit memories locally, sync later
• Backup - Keep a local copy of all memories
• Migration - Import existing MEMORY.md files

Format

# Agent Memory

> Last synced: 2026-02-01T12:00:00Z
> Total memories: 42

---

<!-- id: abc123 -->
<!-- metadata: {"category": "preferences"} -->
User prefers dark mode

---

<!-- id: def456 -->
Another memory here

---

Examples

Store with metadata

agentmemory store "Project deadline is March 15" \
  --category deadline \
  --metadata '{"project": "website-redesign", "priority": "high"}'

Export and backup

# Export to file
agentmemory export --output backup.json --pretty

# Export to stdout (for piping)
agentmemory export > memories-$(date +%Y%m%d).json

Import memories

agentmemory import backup.json

Scripting with JSON output

# Get memories as JSON for processing
agentmemory list --json | jq '.memories[].content'

# Search and process results
agentmemory search "preferences" --json | jq '.[] | .content'

For AI Agents

Add to your agent's skill file:

# Store a memory
agentmemory store "$MEMORY_CONTENT"

# Search before asking
RESULTS=$(agentmemory search "$QUERY" --json)

# Sync at session start
agentmemory sync --pull

Support

• Documentation: https://agentmemory.cloud/docs
• Dashboard: https://agentmemory.cloud/dashboard
• Issues: https://github.com/agentmemory/agentmemory-cli/issues

License

MIT