bitvoker is a notification system that receives raw messages over TCP, filters and processes them through customizable rules and optional AI, then delivers them to over 100 destinations including Slack, Discord, Telegram, Microsoft Teams, Email, and more via Apprise.

bitvoker turns raw text into targeted, intelligent alerts. Send it logs, text, or any data, and configure rules to control exactly what happens. With regex matching and AI processing, here are some examples:

1. Security logs from web-gateway-03 contain Failed login attempt for user admin from an IP outside 192.168.1.0/24, use a local LLM to identify the attack origin and recommend action, then send only the AI summary to the SOC team's Slack channel.

2. A scraped product page contains Sony WH-1000XM5 with a discount over 15%, extract the current price, original price, and buy link using AI, then send the deal to a Telegram chat.

3. Database logs from db-prod-01 show long_query_threshold exceeded over 1000ms, summarize the impact using Meta's LLAMA4, send both the summary and the original log (only if it contains an IP starting with 10.0.0.) to the DBA team's Microsoft Teams channel and email inbox.

• Multi-platform support: notifications for

  and many more thanks to Apprise integration.

• AI Processing: refine messages using customizable pre-prompts with Meta's LLAMA4 or self-hosted Ollama
• Flexible Rule System: regex matching, source filtering, and per-rule AI and destination control
• Web Dashboard: modern interface for configuration, notification history, and log viewing
• Authentication: optional login for the web UI and token verification for TCP messages
• Dynamic Configuration: update settings and rules without restarting the server
• Notification History: browse and filter past notifications with timestamps and source info

bitvoker can optionally process messages with AI before delivery:

• Meta LLAMA4 (default): free cloud-based processing via meta.ai, subject to rate limits
• Ollama: self-hosted local processing via ollama.com, recommended for privacy and reliability

Define pre-prompts in your rules to control how AI processes each message. See the wiki for the full rule reference.

services:
  bitvoker:
    image: ghcr.io/rmfatemi/bitvoker:latest
    container_name: bitvoker
    # host mode recommended (see wiki for details)
    network_mode: host
    # for bridge mode, comment out the line above and uncomment below
    # ports:
    #   - "8083:8083" # TCP server
    #   - "8084:8084" # TLS server
    #   - "8085:8085" # Web UI HTTPS
    #   - "8086:8086" # Web UI HTTP
    volumes:
      - bitvoker_data:/app/data
      - /etc/localtime:/etc/localtime:ro
    environment:
      # Optional: uncomment to enable web UI login
      # - BITVOKER_USERNAME=admin
      # - BITVOKER_PASSWORD=changeme
    restart: unless-stopped

volumes:
  bitvoker_data:
    name: bitvoker_data

docker-compose up -d

Requires Python 3.11+ and GNU Make.

git clone https://github.com/rmfatemi/bitvoker.git
cd bitvoker
make install
make run

Set BITVOKER_USERNAME and BITVOKER_PASSWORD environment variables to enable login for the web UI. When not set, the UI is accessible without authentication.

For TCP messages, optionally configure a message_token in the settings to require authentication:

Without token authentication (default, if message_token is not set):

echo "your message" | nc {server_ip} 8083

With token authentication (if message_token is configured in settings):

echo "TOKEN:your_secret_token:your message" | nc {server_ip} 8083

Messages without the correct token prefix will be rejected when token authentication is enabled.

Send messages to bitvoker over TCP using plaintext (port 8083) or TLS (port 8084).

echo "your notification" | nc {server_ip} 8083

echo "your notification" | openssl s_client -connect {server_ip}:8084

import socket, ssl

context = ssl.create_default_context()
with socket.create_connection(("{server_ip}", 8084)) as sock:
    with context.wrap_socket(sock, server_hostname="{server_ip}") as s:
        s.sendall(b"your notification")

Access the web UI at https://{server_ip}:8085 (or http on port 8086) to configure destinations, rules, AI settings, and view notification history and logs.

This project is licensed under the MIT License - see the LICENSE file for details.