memory-receive

name	memory-receive
description	Receive shared project memory from LAN or relay
user-invocable	true
arguments	[{"name":"passphrase","description":"Shared passphrase for decryption (required)","required":true},{"name":"--relay","description":"Use WebSocket relay server for remote receiving","required":false},{"name":"--room","description":"Room code to join (required with --relay)","required":false}]

Memory Receive

You are receiving project memory shared from another Claude Code instance.

Step 1: Receive the payload

Run the receive.py script with the user's passphrase:

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/receive.py "{{passphrase}}"

This will:

1. Discover the sharing service via mDNS on the local network
2. Authenticate using the passphrase (challenge-response)
3. Decrypt and output the memory payload

Step 2: Process the payload

Check if the received payload is a JSON memory bundle or plain Markdown.

If JSON memory bundle ({"type": "memory_bundle", ...})

Run the install script to save files to the local memory directory:

python3 ${CLAUDE_PLUGIN_ROOT}/skills/memory-receive/scripts/install.py

Feed the received JSON payload via stdin. The script will:

• Save each file to ~/.claude/projects/<project>/memory/
• Back up existing files with .bak extension before overwriting
• Display the list of saved files

After installation, tell the user which files were saved and that the memory is now active.

If plain Markdown (distilled mode)

Display the received memory content to the user in full. Then offer to save it:

• Write the content to ~/.claude/projects/<project>/memory/MEMORY.md (back up existing file first)
• Or let the user decide where to incorporate the information

Relay Mode (Remote Receiving)

If the user passes --relay --room <room_code>, use the relay server:

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/receive.py --relay --room "{{room_code}}" "{{passphrase}}"

This will:

1. Connect to the relay server at wss://relay.fireamulet.com
2. Join the room using the provided room code
3. Authenticate and decrypt the payload over WebSocket
4. Requires pip install websockets if not already installed

Step 3: Confirm to the user

After processing, tell the user:

• What memory was received (file list or distilled summary)
• That the memory is now available in their auto memory directory
• They can review and edit the files in ~/.claude/projects/<project>/memory/