stillrunning 1.3.0

Lightweight process monitor with Telegram alerts and Shield security. Auto-restarts crashed processes, monitors CPU/memory/disk, blocks SSH attacks, sends alerts.

StillRunning

Lightweight process monitor with auto-restart, Telegram alerts, and live dashboard. Watches your processes, restarts crashes, monitors system resources, and sends you alerts.

How It Works

1. Pay at stillrunning.io — Basic ($29/mo) or AI tier ($49/mo)
2. Get your token via email — arrives instantly after checkout
3. Run one command on your server:

   curl -sSL https://stillrunning.io/install | python3 - --token YOUR_TOKEN
   

4. Dashboard goes live — yourname.stillrunning.io shows all your processes within 60 seconds

Features

• Process Watchdog: Auto-restarts crashed screen sessions, systemd services, PM2 processes
• Live Dashboard: See all your processes at yourname.stillrunning.io
• Resource Monitoring: CPU, memory, disk usage alerts
• Per-Process Memory: Alerts when individual processes exceed thresholds
• Log Rotation: Auto-archives logs when they exceed size limits
• Health File Monitoring: Alerts when your app stops updating its heartbeat file
• Telegram Alerts: Get notified instantly when something breaks
• Telegram Commands: Text "status" to get a live summary
• Daily Heartbeat: Receive a daily health summary
• Restart Limits: Disables processes after 3 consecutive failures to prevent restart loops

AI Tier Adds

• AI Crash Diagnosis: Claude analyzes crash logs and explains what went wrong
• Fix Suggestions: Get code fix suggestions sent to Telegram with approve/reject buttons
• Daily Health Audit: Automated code quality and infrastructure review every morning
• Pattern Detection: Learns from your crash history to predict and prevent issues

Shield — Intelligent Attack Protection

Shield monitors SSH brute force attacks and automatically blocks malicious IPs.

Basic Tier (included free)

• SSH failed login counter (reads /var/log/auth.log)
• Automatic fail2ban integration after 10 failed attempts
• Security grade A/B/C/D based on attack volume
• Daily security summary in Telegram heartbeat

AI Tier (intelligent protection)

• Intelligent threat scoring: LOW, MEDIUM, HIGH, REPEAT offender detection
• Tiered punishment: fail2ban for LOW, iptables 24h for MEDIUM, permanent ban for HIGH/REPEAT
• AbuseIPDB reporting for HIGH+ threats (requires API key)
• Claude Haiku pattern analysis every 50 attacks
• Shared threat network: every customer protects every other customer
• Pre-emptive blocking from shared blocklist

Multi-Server Support

Monitor multiple servers from one account. Each server gets its own token and appears on your unified dashboard.

Basic tier: Up to 3 servers AI tier: Unlimited servers

Add a new server to your account:

stillrunning --add-server --token YOUR_CUSTOMER_TOKEN --name my-new-server

This generates a new server token. Run the install command with that token on your new server.

Telegram Two-Way Control (AI tier)

AI tier customers can text their Telegram bot to control their servers:

Commands:

• status — full system status in plain English
• restart [process] — restart a named process (must be in config)
• logs [process] — view last 20 lines of process log (sanitized)
• shield — Shield security summary
• help — list available commands

AI Chat (AI tier only): Any non-command message routes to Claude Haiku with full system context. Ask questions in plain English:

• "Why did my API server crash?"
• "Is my disk usage concerning?"
• "What should I check first?"

Rate limited to 20 messages per hour. All log content is sanitized before sending to AI.

Shield Security — Log Sanitization

When AI crash diagnosis analyzes your logs, all potential secrets are automatically redacted:

• API keys (Anthropic, Stripe, Bearer tokens)
• Passwords and secrets in config format
• Long hex strings (common secret format)

This happens automatically — you don't need to configure anything.

Quick Start

curl -sSL https://stillrunning.io/install | python3 - --token YOUR_TOKEN

The installer will:

1. Validate your token with stillrunning.io
2. Auto-detect running screen sessions, systemd services, and PM2 processes
3. Generate stillrunning.yaml in your current directory
4. Start pushing status to your dashboard every 30 seconds

Self-Hosted (Free)

If you prefer to self-host without a dashboard:

pip install stillrunning
stillrunning --setup

The setup wizard will:

1. Scan for running screen sessions and systemd services
2. Detect log files in your working directory
3. Ask for your app name and Telegram credentials
4. Generate stillrunning.yaml automatically
5. Start monitoring

Manual Configuration

Create stillrunning.yaml:

app_name: "MyApp"
working_dir: "/home/user/myapp"

telegram_bot_token: "123456789:ABCdefGHI..."
telegram_chat_id: "987654321"

processes:
  - name: "api"
    screen: "api"
    script: "api.py"
  - name: "worker"
    screen: "worker"
    script: "worker.py"

log_files:
  - path: "app.log"
    max_mb: 10
    keep_archives: 5

health_file: "status.json"
health_max_age_sec: 180

thresholds:
  cpu_percent: 85
  mem_percent: 85
  disk_percent: 85
  process_mem_mb: 500

Run:

screen -dmS stillrunning stillrunning

Telegram Setup

1. Create a bot: Message @BotFather and send /newbot
2. Copy the bot token
3. Get your chat ID: Message @userinfobot and send /start
4. Add both to your config

Telegram Commands

Text these to your bot:

• status - Get live process and resource status
• help - Show available commands
• enable all - Re-enable disabled processes

Configuration Reference

Field	Default	Description
app_name	"StillRunning"	Name shown in alerts
working_dir	current dir	Where your scripts live
telegram_bot_token	-	From @BotFather
telegram_chat_id	-	Your Telegram user ID
processes	[]	List of processes to monitor
log_files	[]	List of logs to auto-rotate
health_file	null	File to monitor for freshness
health_max_age_sec	180	Alert if health file older than this
thresholds.cpu_percent	85	CPU alert threshold
thresholds.mem_percent	85	Memory alert threshold
thresholds.disk_percent	85	Disk alert threshold
thresholds.process_mem_mb	500	Per-process memory threshold
intervals.process_check_sec	30	How often to check processes
intervals.resource_check_sec	60	How often to check resources
intervals.heartbeat_sec	86400	Daily heartbeat interval
restart_cooldown_sec	120	Minimum time between restarts
max_consecutive_failures	3	Disable process after this many failures

Requirements

• Linux (uses /proc/stat, /proc/meminfo, screen, pgrep)
• Python 3.10+
• PyYAML

License

MIT