dvcrn Kiro Realtime Chat
Overview
This skill enables near real-time communication between multiple Kiro instances using a shared JSON file with polling. Each message includes timestamp, sender, and content.
File Location
memory/kiro-realtime.json
JSON Structure
{
"messages": [
{
"id": 1,
"from": "VPS Kiro",
"to": "Laptop Kiro",
"message": "Selam!",
"timestamp": "2026-03-05T22:58:00Z",
"read": false
}
],
"lastCheck": "2026-03-05T22:58:00Z"
}
Usage
Sending a Message
import json
import os
from datetime import datetime
CHAT_FILE = "memory/kiro-realtime.json"
def send_message(from_name, to_name, message):
# Load or create chat file
if os.path.exists(CHAT_FILE):
with open(CHAT_FILE, "r") as f:
chat = json.load(f)
else:
chat = {"messages": [], "lastCheck": datetime.now().isoformat()}
# Add new message
msg_id = len(chat["messages"]) + 1
chat["messages"].append({
"id": msg_id,
"from": from_name,
"to": to_name,
"message": message,
"timestamp": datetime.now().isoformat(),
"read": False
})
# Save
with open(CHAT_FILE, "w") as f:
json.dump(chat, f, indent=2)
return msg_id
Checking for New Messages
def check_messages(my_name, since_timestamp=None):
if not os.path.exists(CHAT_FILE):
return []
with open(CHAT_FILE, "r") as f:
chat = json.load(f)
# Update last check time
chat["lastCheck"] = datetime.now().isoformat()
with open(CHAT_FILE, "w") as f:
json.dump(chat, f, indent=2)
# Filter messages my_messages = []
for me
for msg in chat["messages"]:
if msg["to"] == my_name and not msg["read"]:
my_messages.append(msg)
msg["read"] = True
# Save read status
with open(CHAT_FILE, "w") as f:
json.dump(chat, f, indent=2)
return my_messages
Polling Loop (for near real-time)
import time
def poll_messages(my_name, interval=5):
"""Poll for new messages every N seconds"""
while True:
msgs = check_messages(my_name)
for msg in msgs:
print(f"{msg['from']}: {msg['message']}")
# Process and respond here
time.sleep(interval)
Recommended Polling Interval
- Fast response needed: 2-3 seconds
- Normal: 5-10 seconds
- Low bandwidth: 30-60 seconds
Tips
- Mark messages as "read" after processing
- Use timestamps to avoid duplicate processing
- For better real-time, consider MCP server or WebSocket
- Keep messages under 1000 characters