GitHub - maqalaqil/mhost: The AI-powered process manager.

                 ██╗                    ██╗
 ██████╗██████╗ ██████╗  ██████╗ ███████╗████████╗
██╔████╗██╔══██╗██╔══██╗██╔═══██╗██╔════╝╚══██╔══╝
██║██╔██║██████╔╝██║  ██║██║   ██║███████╗   ██║
██║╚████║██╔══██╗██║  ██║██║   ██║╚════██║   ██║
╚██╗╚██╗██║  ██║██████╔╝╚██████╔╝███████║   ██║
 ╚══╝╚═╝╚═╝  ╚═╝╚═════╝  ╚═════╝ ╚══════╝   ╚═╝

The PM2 alternative that PM2 should have been.

Built in Rust. Single binary. Zero runtime dependencies. Built-in AI agent.

🌐 mhostai.com · Pricing · Docs · Playground

Why mhost?

	PM2	mhost
Runtime	Requires Node.js	Single 14MB binary
Health checks	Alive/dead only	HTTP, TCP, script probes
Log search	Grep files manually	Built-in FTS5 full-text search
Notifications	Plugin required	Telegram, Slack, Discord, Email, PagerDuty, Teams, Ntfy, Webhook
Metrics	Basic	Prometheus export, time-series, alerts, auto-remediation
Proxy	None	Built-in reverse proxy with auto-TLS
Deploy	Basic	Git pull + hooks + rollback with history
Dashboard	Web only	Terminal TUI with sparklines
Groups	None	Dependency ordering with topological sort
Restart	Basic	Exponential backoff + circuit breaker
Config	JS only	TOML, YAML, JSON
AI	None	Built-in LLM intelligence (OpenAI/Claude) — diagnose, optimize, ask
Autonomous agent	None	AI agent with LLM tool calling — monitors, acts, reports to Telegram
Cloud fleet	None	SSH fleet management — AWS, Azure, DO, Railway auto-import
Zero-downtime	None	Rolling reload with health-check gate
Dev mode	None	Built-in file watcher + auto-restart (replaces nodemon)
Web dashboard	Paid	Free built-in web UI on localhost
Load testing	None	Built-in HTTP benchmark tool
Canary deploys	None	Canary scale + monitor + auto-promote/rollback
Snapshots	None	Capture & restore full process state
SSL monitor	None	Certificate expiry checking
SLA reports	None	Uptime reports with target tracking
Migration	None	Auto-convert PM2 configs to mhost

Installation

Homebrew	brew install maqalaqil/tap/mhost
npm	npm install -g @maqalaqil93/mhost
Cargo	cargo install mhost
curl	curl -fsSL https://mhostai.com/install.sh \| sh
PowerShell	irm https://mhostai.com/install.ps1 \| iex
Scoop	scoop install mhost
From source	git clone https://github.com/maqalaqil/mhost && cd mhost && cargo build --release

Quick Start

Start any process

# Node.js
mhost start "node server.js" --name api

# Python
mhost start "python3 worker.py" --name worker

# Any binary
mhost start "./my-service --port 8080" --name service

# From ecosystem config
mhost start mhost.toml

See what's running

$ mhost list

id   name          status    pid     inst  uptime      restarts  memory
────────────────────────────────────────────────────────────────────────
a1b2 api-server    online    12345   0     2d 14h 22m  0         128MB
c3d4 worker        online    12346   0     2d 14h 22m  0         64MB
e5f6 worker        online    12347   1     2d 14h 22m  0         62MB
g7h8 static-site   online    12348   0     1d 8h 15m   1         32MB

Manage processes

mhost stop api-server        # Graceful SIGTERM -> wait -> SIGKILL
mhost restart worker         # Zero-downtime restart
mhost scale worker 4         # Scale to 4 instances
mhost delete api-server      # Remove from registry
mhost stop all               # Stop everything

Persist across reboots

mhost save                   # Save current process list
mhost resurrect              # Restore after daemon restart
mhost startup                # Auto-start mhost on boot (launchd/systemd)

Ecosystem Config

Define your entire stack in one file. Supports TOML, YAML, and JSON.

# mhost.toml

# ─── API Server ──────────────────────────────────────────────
[process.api-server]
command = "node server.js"
cwd = "/app/api"
env = { NODE_ENV = "production", PORT = "3000" }
instances = 4                         # Run 4 instances (cluster mode)
max_memory = "512MB"                  # Kill & restart if exceeds
max_restarts = 15                     # Circuit breaker threshold
min_uptime = "2s"                     # Crash = exit before this
restart_delay = "100ms"               # Base delay (exponential backoff)
grace_period = "5s"                   # SIGTERM wait before SIGKILL

[process.api-server.health.http]      # Health probe
url = "http://localhost:3000/health"
interval = "10s"
timeout = "3s"
retries = 3

# ─── Background Worker ──────────────────────────────────────
[process.worker]
command = "python3 worker.py"
cwd = "/app/worker"
instances = 2
max_restarts = 10

# ─── Scheduled Cleanup ──────────────────────────────────────
[process.cleanup]
command = "sh cleanup.sh"
cwd = "/app/scripts"
cron_restart = "0 3 * * *"           # Restart daily at 3am

# ─── Process Groups (dependency ordering) ────────────────────
[groups.backend]
depends_on = []
processes = ["api-server", "worker"]

[groups.frontend]
depends_on = ["backend"]              # Starts AFTER backend group
processes = ["static-site"]

# ─── Notifications ───────────────────────────────────────────
[notifications.telegram]
type = "telegram"
bot_token = "${MHOST_TELEGRAM_TOKEN}"
chat_id = "${MHOST_TELEGRAM_CHAT}"
events = ["crash", "errored", "health_fail", "5xx_error"]
throttle = "60s"

[notifications.slack]
type = "slack"
webhook = "${SLACK_WEBHOOK_URL}"
events = ["crash", "deploy_success", "deploy_fail"]

# ─── Log Sinks ───────────────────────────────────────────────
[logs.sinks.graylog]
type = "gelf"
host = "graylog.local"
port = 12201
processes = "api-*"

[logs.sinks.loki]
type = "loki"
url = "http://loki.local:3100/loki/api/v1/push"
labels = { env = "production" }

# ─── Alerts & Auto-Remediation ───────────────────────────────
[alerts.high-memory]
process = "api-server"
condition = "memory > 450MB for 5m"
notify = ["telegram", "slack"]
action = "restart"

[remediation.zombie-detection]
condition = "cpu < 1% AND health_fail for 5m"
action = "restart"
cooldown = "10m"
notify = ["slack"]

# ─── Deploy ──────────────────────────────────────────────────
[deploy.production]
repo = "git@github.com:user/app.git"
branch = "main"
path = "/var/www/app"
pre_deploy = ["npm install", "npm run build"]
post_deploy = ["mhost restart api-server"]

# ─── Reverse Proxy ───────────────────────────────────────────
[proxy]
listen = "0.0.0.0:80"
tls_listen = "0.0.0.0:443"
acme = true
acme_email = "admin@example.com"

[proxy.routes]
"api.example.com" = { target = "api-server", port = 3000 }
"app.example.com" = { target = "static-site", port = 8080 }

mhost start mhost.toml

Health Probes

Processes only transition to online after health checks pass. Failures trigger restarts.

# HTTP probe — check status code
[process.api.health.http]
url = "http://localhost:3000/health"
interval = "10s"
timeout = "3s"
retries = 3

# TCP probe — check port is open
[process.db.health.tcp]
host = "127.0.0.1"
port = 5432
interval = "5s"

# Script probe — check exit code
[process.app.health.script]
command = "./check-health.sh"
interval = "15s"

mhost health api-server       # Show health status per instance

Process Groups & Dependencies

Start services in dependency order. Stop in reverse.

[groups.database]
processes = ["postgres", "redis"]

[groups.backend]
depends_on = ["database"]              # postgres & redis start first
processes = ["api-server", "worker"]

[groups.frontend]
depends_on = ["backend"]               # api-server starts before nginx
processes = ["nginx"]

mhost start --group backend            # Starts database group first, then backend
mhost stop --group frontend            # Stops in reverse dependency order

Auto-Restart & Circuit Breaker

Process crashes
    |
    v
Restart attempt #1 (delay: 100ms)
    |
    v
Restart attempt #2 (delay: 200ms)     # Exponential backoff
    |
    v
Restart attempt #3 (delay: 400ms)     # 100ms * 2^attempt
    |                                   # Capped at 30 seconds
    ...
    v
Attempt #15 within min_uptime
    |
    v
ERRORED (circuit breaker open)         # Stops retrying
                                        # Sends notification

Notifications

Quick Setup (CLI)

mhost notify setup                    # Interactive wizard

  mhost Notification Setup

  Select channel type:
    1) Telegram
    2) Slack
    3) Discord
    4) Generic Webhook

  Channel type (1-4): 1

  Telegram Setup
  ─────────────────────────────
  1. Message @BotFather on Telegram
  2. Send /newbot and follow the instructions
  3. Copy the bot token below
  4. Message your bot, then get your chat ID from @userinfobot

  Bot token: 123456:ABC-DEF...
  Chat ID: 987654321
  Channel name [telegram]: telegram

  Available alert events:
    1   crash
    2   restart
    3   errored
    4   stopped
    5   recovered
    6   health_fail
    7   high_restarts
    8   5xx_error
    9   oom_kill
    10  deploy_success
    11  deploy_fail
    *   All events

  Select events (comma-separated numbers, or * for all): *

  ✓ Channel 'telegram' configured and saved

Manage Channels

mhost notify list                     # Show all configured channels
mhost notify test telegram            # Send a test message
mhost notify enable telegram          # Enable a channel
mhost notify disable telegram         # Disable without removing
mhost notify remove telegram          # Delete a channel
mhost notify events                   # Show all event types
mhost notify events telegram          # Show channel subscriptions
mhost notify start                    # Launch notifier as managed process

Supported Channels

Channel	Transport	Features
Telegram	Bot API	Rich markdown, inline buttons
Slack	Webhook	Block Kit, color-coded
Discord	Webhook	Embeds, severity colors
Webhook	HTTP POST	Custom headers, HMAC-SHA256 signing
Email	SMTP/TLS	HTML templates, digest mode
PagerDuty	Events API v2	Severity mapping, auto-resolve
Microsoft Teams	Webhook	Adaptive cards
Ntfy	HTTP	Self-hosted push notifications

Alert Events

Event	Trigger
`crash`	Process exited with non-zero code
`restart`	Process auto-restarted by mhost
`errored`	Max restarts exceeded (circuit breaker tripped)
`stopped`	Process was stopped
`recovered`	Process came back online after failure
`health_fail`	Health check probe failed
`high_restarts`	Process restarted 5+ times
`5xx_error`	Health endpoint returned HTTP 5xx
`oom_kill`	Process killed for exceeding memory limit
`deploy_success`	Deploy completed successfully
`deploy_fail`	Deploy failed

Throttling & Escalation

[notifications.slack]
throttle = "60s"                       # Suppress duplicate alerts for 60s

[notifications.escalation]
chain = ["slack", "telegram", "pagerduty"]
escalate_after = "5m"                  # If no ack in 5min, notify next channel

Log Engine

Built-in full-text search (SQLite FTS5)

# Search across all logs
mhost logs api --search "connection refused" --since 1h

# Structured queries
mhost logs api --where "level=error AND status>=500"

# Regex filtering
mhost logs api --grep "status=[45]\d\d"

# Aggregation
mhost logs --all --where "level=error" --since 1h --count-by process

# Export
mhost logs api --since 7d --format jsonl > export.jsonl

JSON auto-detection

If your process outputs JSON to stdout, mhost automatically parses and indexes every field:

{"level":"error","message":"Connection timeout","status":503,"latency_ms":5032}

All fields become searchable: --where "status>=500", --search "timeout".

External Log Sinks

Push logs to your existing infrastructure:

[logs.sinks.graylog]
type = "gelf"
host = "graylog.local"
port = 12201
transport = "udp"
processes = "api-*"                    # Glob pattern matching

[logs.sinks.loki]
type = "loki"
url = "http://loki.local:3100/loki/api/v1/push"

[logs.sinks.elasticsearch]
type = "elasticsearch"
url = "http://es.local:9200"
index = "mhost-logs-{date}"           # Date-templated index names

[logs.sinks.syslog]
type = "syslog"
host = "syslog.local"
port = 514

Retention Policies

info  logs  ->  7 days
warn  logs  ->  30 days
error logs  ->  30 days
fatal logs  ->  90 days

Metrics & Prometheus

CLI

mhost metrics show api-server         # Current CPU, memory, uptime
mhost metrics history api --metric cpu --since 24h
mhost metrics start --listen 0.0.0.0:9090  # Start Prometheus exporter

Prometheus endpoint

GET http://localhost:9090/metrics

# HELP mhost_process_cpu_percent CPU usage percentage
# TYPE mhost_process_cpu_percent gauge
mhost_process_cpu_percent{name="api",instance="0"} 42.5
mhost_process_memory_bytes{name="api",instance="0"} 134217728
mhost_process_uptime_seconds{name="api",instance="0"} 86400
mhost_process_restart_total{name="api",instance="0"} 2

Alert Rules & Auto-Remediation

[alerts.high-memory]
process = "api-server"
condition = "memory > 450MB for 5m"
notify = ["telegram", "slack"]
action = "restart"                     # Auto-restart on breach

[remediation.zombie-detection]
condition = "cpu < 1% AND health_fail for 5m"
action = "restart"
cooldown = "10m"                       # Don't re-trigger for 10 min

Reverse Proxy

Built-in HTTP/HTTPS reverse proxy with virtual host routing.

[proxy]
listen = "0.0.0.0:80"
tls_listen = "0.0.0.0:443"
acme = true                            # Auto-TLS via Let's Encrypt
acme_email = "admin@example.com"

[proxy.routes]
"api.example.com" = { target = "api-server", port = 3000, strategy = "least_connections" }
"app.example.com" = { target = "frontend", port = 8080, sticky = true }

Features: Load balancing (round-robin, least-connections, IP-hash), sticky sessions, WebSocket passthrough, self-signed TLS for local dev, ACME auto-cert for production.

mhost proxy                            # Show current routes

Deploy Engine

mhost deploy production                # git pull + hooks + graceful reload
mhost rollback production              # Revert to previous successful deploy

[deploy.production]
repo = "git@github.com:user/app.git"
branch = "main"
path = "/var/www/app"
pre_deploy = ["npm install", "npm run build"]
post_deploy = ["mhost restart api-server"]

Deploy history is tracked in SQLite with commit hashes, timestamps, and status.

Cloud Fleet Management

Manage processes across any number of remote servers from your local terminal via SSH.

Add Servers

mhost cloud add prod-api --host 54.123.45.67 --key ~/.ssh/id_rsa
mhost cloud add staging --host staging.myapp.com --user deploy

Auto-Import from Cloud Providers

mhost cloud import aws --region us-east-1 --tag env=production
mhost cloud import digitalocean --tag production
mhost cloud import azure --resource-group myapp
mhost cloud import railway --project myapp

Remote Operations

mhost cloud list                          # Show all servers
mhost cloud status                        # Health check entire fleet
mhost cloud deploy prod-api mhost.toml    # Deploy config to remote
mhost cloud exec prod-api "mhost list"    # Run command on remote
mhost cloud logs prod-api api-server      # Stream remote logs
mhost cloud restart all api-server        # Restart across ALL servers
mhost cloud scale prod-api worker 4       # Scale on specific server
mhost cloud sync mhost.toml              # Push config to all servers
mhost cloud ssh prod-api                  # Open SSH shell
mhost cloud install prod-api              # Install mhost on remote
mhost cloud update all                    # Update mhost on all servers

AI-Powered Cloud Operations

mhost cloud ai-setup "I need a Node API on EC2 with 2 workers"
mhost cloud ai-diagnose prod-api          # AI analyzes remote server
mhost cloud ai-migrate staging prod-api   # AI plans migration

Fleet Config (`~/.mhost/fleet.json`)

{
  "servers": {
    "prod-api": {
      "host": "54.123.45.67",
      "port": 22,
      "user": "deploy",
      "auth": "key",
      "key_path": "~/.ssh/id_rsa",
      "tags": ["production", "api"],
      "provider": "aws",
      "region": "us-east-1"
    }
  },
  "groups": {
    "production": ["prod-api", "prod-worker"]
  }
}

Supports: AWS EC2, Azure VMs, DigitalOcean, Railway, any SSH-accessible server.

Cloud-Native Provider Integrations ^BETA

Direct API integration with 10 cloud providers — no SSH required.

Supported Providers

Quick Start

mhost cloud auth railway              # Configure credentials
mhost cloud provision --provider railway --name api --type web --image node:20 --port 3000
mhost cloud services                  # List all services across providers
mhost cloud scale api 4               # Scale to 4 instances
mhost cloud cost                      # See spending across providers
mhost cloud drift                     # Check for config drift
mhost cloud export terraform          # Generate Terraform files

Secret Management

mhost cloud secrets set api DATABASE_URL "postgres://..."
mhost cloud secrets list api
mhost cloud secrets remove api OLD_KEY

Backups & Export

mhost cloud backup api                # Backup service data
mhost cloud backup-list               # List all backups
mhost cloud export terraform          # Generate Terraform HCL
mhost cloud export docker-compose     # Generate docker-compose.yml
mhost cloud export kubernetes         # Generate K8s manifests

Chat Bot Control

Control your processes from anywhere using a Telegram or Discord bot. Send /status, /start, /stop, and more — directly from your phone.

Setup

mhost bot setup

  mhost Bot Setup

  Select platform:
    1) Telegram
    2) Discord

  Platform (1-2): 1
  Bot token: 123456:ABC-DEF...
  Your user/chat ID (admin): 987654321

  ✓ Bot configured for telegram
  Config: ~/.mhost/bot.json
  Start:  mhost bot enable

Chat Commands

Command	Permission	Description
`/status`	Viewer+	List all processes with status
`/start <app>`	Operator+	Start a process
`/stop <app>`	Operator+	Stop a process
`/restart <app>`	Operator+	Restart a process
`/logs <app>`	Viewer+	Show recent log lines
`/health <app>`	Viewer+	Show health check status
`/scale <app> <N>`	Operator+	Scale to N instances
`/deploy <env>`	Operator+	Trigger a deploy
`/ai <question>`	Admin	Ask the AI assistant
`/help`	All	Show available commands

Permission System

Three roles control who can do what:

Role	Commands	Assigned by
Admin	All commands including AI and permission management	First user during setup
Operator	start, stop, restart, scale, logs, health, deploy	Admin
Viewer	status, logs, health, help (read-only)	Admin

Users not in any list are denied all commands. Users can be explicitly blocked.

mhost bot add-admin  987654321
mhost bot add-operator 111222333
mhost bot add-viewer  444555666
mhost bot remove-user 444555666

CLI Commands

Command	Description
`mhost bot setup`	Interactive setup wizard (Telegram/Discord)
`mhost bot enable`	Start the bot (long-polling)
`mhost bot disable`	Stop the bot
`mhost bot status`	Show platform, enabled state, user lists
`mhost bot permissions`	Show role → user mapping
`mhost bot add-admin <id>`	Grant admin access
`mhost bot add-operator <id>`	Grant operator access
`mhost bot add-viewer <id>`	Grant viewer access
`mhost bot remove-user <id>`	Remove user from all roles
`mhost bot logs`	Show recent bot audit log (last 20 entries)

Security Features

Role-based access control — three tiers with explicit deny lists
Confirmation prompts — destructive commands (stop, restart) require /confirm before executing
Rate limiting — configurable commands-per-minute cap per user (default: 30/min)
Audit logging — every command is logged to ~/.mhost/bot-audit.jsonl with user ID, timestamp, and result
Auto-alerts — daemon crash/recover events forwarded to admin users automatically

Bot Config (`~/.mhost/bot.json`)

{
  "enabled": true,
  "platform": "telegram",
  "token": "123456:ABC-DEF...",
  "permissions": {
    "admins": [987654321],
    "operators": [111222333],
    "viewers": [444555666],
    "blocked": []
  },
  "confirm_destructive": true,
  "auto_alerts": true,
  "rate_limit": 30
}

AI Intelligence

The first process manager with built-in LLM capabilities. Supports OpenAI (GPT-4o) and Claude (Sonnet/Opus).

Setup

mhost ai setup

  mhost AI Setup

  Select LLM provider:
    1) OpenAI (GPT-4o, GPT-4o-mini)
    2) Claude (Sonnet, Haiku, Opus)

  Provider (1-2): 1
  API key: sk-...
  Model [gpt-4o]: gpt-4o

  ✓ AI configured with openai (gpt-4o)

Or configure via file (~/.mhost/ai.json) or environment variables.

Crash Diagnosis

mhost ai diagnose api-server

  Analyzing crash for 'api-server'...

  ## Root Cause
  The process crashed due to an unhandled promise rejection in database.js:42.
  The connection pool was exhausted after 15 concurrent requests exceeded the
  pool limit of 10.

  ## Impact
  Severity: HIGH — All API requests failed for 12 seconds until auto-restart.

  ## Fix Steps
  1. Increase pool size: `max_connections: 25` in database config
  2. Add connection timeout: `idle_timeout: 30000`

  ## Prevention
  - Add health check for DB connection pool utilization
  - Set up alert: `condition = "memory > 256MB for 5m"`

  ## Config Suggestions
  max_memory = "512MB"    # Currently 256MB — too tight
  max_restarts = 20       # Currently 15 — increase headroom

Natural Language Log Queries

mhost ai logs api "show me all timeout errors in the last hour"
mhost ai logs worker "what errors happened during the deploy?"
mhost ai logs api "count errors by type today"

Generate Config from Description

mhost ai config "I have a Node.js API on port 3000 with 2 Python celery workers
                  and a React frontend. Add health checks and process groups."

Outputs a complete, valid mhost.toml ready to use.

Performance Optimization

mhost ai optimize api-server

Analyzes CPU/memory trends and suggests instance count, memory limits, restart thresholds.

Incident Post-Mortem

mhost ai postmortem api-server

Generates a full Markdown incident report with timeline, root cause, impact, and action items.

Ask Anything

mhost ai ask "which process is using the most memory?"
mhost ai ask "why has my worker restarted 12 times today?"
mhost ai ask "should I scale up my API?"

More AI Commands

mhost ai watch                        # Scan all processes for anomalies
mhost ai explain mhost.toml           # Explain config in plain English
mhost ai suggest                      # Get proactive improvement suggestions

AI Config

# In mhost.toml
[ai]
provider = "openai"                    # or "claude"
api_key = "${OPENAI_API_KEY}"          # env var expansion
model = "gpt-4o"                       # any supported model
max_tokens = 4096

Autonomous Agent

An AI agent that continuously monitors your processes, diagnoses issues, and takes action — communicating through Telegram.

Setup

mhost agent setup     # Configure LLM, Telegram, autonomy level
mhost agent start     # Start the agent
mhost agent stop      # Stop
mhost agent status    # Show config

How It Works

The agent runs as a managed mhost process. Every 30 seconds it:

Observes — checks all process statuses and logs
Thinks — sends observations to the LLM (GPT-4o/Claude) with tool definitions
Acts — executes mhost commands (restart, scale, etc.) via tool calling
Reports — sends findings and actions to your Telegram

Conversation

Chat naturally with the agent via Telegram:

You: "my api keeps crashing, can you check?"
Agent: "Checked api-server logs — EADDRINUSE on port 3000.
        Previous instance didn't release the port.
        I've restarted it. Now online (PID 45123)."

You: "scale workers to 4"
Agent: "Done. Workers scaled to 4 instances. All online."

You: "what's using the most memory?"
Agent: "api-server: 128MB. Workers: 64MB each. Total: 320MB."

Autonomy Levels

Level	Behavior
`autonomous`	Acts on its own, notifies you after
`supervised`	Proposes actions, waits for your approval
`manual`	Only acts when you send a message

Safety

Blocked actions configurable (delete, kill disabled by default)
Rate limit: 20 actions/hour
Conversation history capped
Full audit trail

Self-Healing Brain

The agent's persistent memory that learns from every incident.

CLI

mhost brain status              # Fleet health scores (0-100 per process)
mhost brain history             # Past incidents and what was done
mhost brain playbooks           # Healing rules (built-in + auto-learned)
mhost brain explain <process>   # Why a process has its health score

How It Works

Process crashes
    │
    ├─ Known pattern? → Apply playbook instantly (no LLM call)
    │   e.g. EADDRINUSE → wait 5s → restart
    │
    ├─ Similar past incident? → Apply same fix
    │
    └─ Unknown? → Call LLM with memory context → Learn from result

Built-in Playbooks

Trigger	Action	Description
`EADDRINUSE`	wait + restart	Port in use — wait for release
`out of memory`	restart	OOM — restart immediately
`Cannot find module`	notify	Bad deploy — needs manual fix
`ECONNREFUSED`	wait + restart	Dependency down — retry
`EACCES`	notify	Permission error
Crash loop (3+ in 10min)	stop + escalate	Too unstable — stop and alert
Memory rising trend	restart	Preemptive leak fix

New playbooks are auto-learned from successful fixes.

Health Scores

mhost brain status

  mhost Brain — Fleet Health
  ──────────────────────────────────────────────────
  api-server           ████████░░ 82/100
  worker               ██████████ 100/100
  monitor              █████░░░░░ 45/100

Zero-Downtime Reload

mhost reload api-server

Starts new instances, waits for health checks to pass, then kills old instances. No dropped requests.

Old instances running (PID 1234, 1235)
    │
    ├── Start new instances
    ├── Wait for health check (up to 30s)
    ├── Health OK → kill old instances
    │   └── ✓ Zero-downtime reload complete
    └── Health FAIL → kill new, keep old
        └── ✗ Reload failed, old instances preserved

Falls back to regular restart if no health check is configured.

Dev Mode

Development mode with file watching and auto-restart — replaces nodemon + dotenv.

mhost dev server.js                        # Watch + auto-restart
mhost dev server.js --watch src/           # Watch specific directory
mhost dev server.js --ext js,ts,json       # Watch specific extensions
mhost dev server.js --env .env.local       # Custom .env file

Web Dashboard

Browser-based monitoring UI accessible from any device.

mhost dashboard                            # Start on http://localhost:9400
mhost dashboard --port 8080                # Custom port

Real-time process cards, analytics strip, live log streaming via SSE, restart/stop/scale buttons.

Load Testing

Built-in HTTP load testing — no external tools needed.

mhost bench https://api.example.com                    # Default: 10s, 10 workers
mhost bench https://api.example.com --duration 30       # 30 second test
mhost bench https://api.example.com --concurrency 50    # 50 concurrent workers

Reports total requests, RPS, average/p99 latency, and error rate.

Canary Deployments

Scale up a canary, monitor for errors, auto-promote or rollback.

mhost canary api-server                               # Default: 10%, 300s
mhost canary api-server --percent 20 --duration 600   # 20% traffic, 10min watch

Snapshots

Capture and restore full process state.

mhost snapshot create before-deploy        # Save all process state
mhost snapshot list                        # Show saved snapshots
mhost snapshot restore before-deploy       # Restore all processes

SSL Certificate Monitoring

mhost certs                                # Check localhost:443
mhost certs --url https://api.example.com --url https://app.example.com

Warns when certificates expire within 30 days.

SLA Uptime Reports

mhost sla api-server                       # Default target: 99.9%
mhost sla api-server --target 99.99        # Four-nines target

Calculates uptime from brain incident data, shows allowed/actual downtime, and budget remaining.

PM2 Migration

mhost migrate --from pm2                   # Auto-convert ~/.pm2/dump.pm2 → mhost.toml

Reads your PM2 dump file and generates a valid mhost ecosystem config with all process settings preserved.

More Tools

mhost replay api-server                    # Replay an incident timeline
mhost replay api-server --time "3:47am"    # Filter around a time
mhost link                                 # Show process dependency graph
mhost cost                                 # Estimate cloud costs from memory usage
mhost diff staging production              # Compare two fleet environments
mhost share api-server                     # Expose via tunnel (ngrok/cloudflared/bore)
mhost run deploy-recipe.txt                # Run a recipe of mhost commands
mhost playground                           # Interactive tutorial

TUI Dashboard

mhost monit

┌─ mhost ─────────────────────────────────────────────────────┐
│ [Processes]  Logs   Metrics   Proxy                         │
├─────────────────────────────────────────────────────────────┤
│ #  Name          Status   PID    CPU%   Memory   Uptime    │
│ ─────────────────────────────────────────────────────────── │
│ 0  api-server    online   12345  12.3%  128MB    2d 14h    │
│ 1  worker        online   12346   3.1%   64MB    2d 14h    │
│ 2  worker        online   12347   2.8%   62MB    2d 14h    │
│ 3  cleanup       online   12348   0.1%   16MB    8h 22m    │
│                                                             │
│ CPU  ▁▂▃▅▇▅▃▂▁▂▃▄▅▃▂▁  12.3%                              │
│ MEM  ▃▃▃▃▃▃▃▄▄▄▃▃▃▃▃▃  128MB                              │
└─────────────────────────────────────────────────────────────┘
  j/k: navigate  Tab: switch view  q: quit  /: search

Keyboard: j/k navigate, g/G top/bottom, Tab switch tabs, / search, q quit, r restart, s stop selected process.

Public API

REST API with bearer token auth, WebSocket real-time streaming, and outbound webhooks.

Quick Start

# Create an API token
mhost api token create --name my-app --role operator

# Use it
curl -H "Authorization: Bearer mhost_tok_..." http://localhost:19516/api/v1/processes

# WebSocket
wscat -c "ws://localhost:19516/api/v1/ws?token=mhost_tok_..."

REST Endpoints (25 total)

Category	Endpoints	Auth
Processes	GET/POST/DELETE /processes/*	viewer/operator
Logs	GET /logs/:name, /logs/:name/search	viewer
Health	GET /health, /health/:name	public/viewer
Metrics	GET /metrics, /metrics/:name	viewer
System	POST /save, /resurrect, /kill	operator/admin
Tokens	GET/POST/DELETE /tokens/*	admin
Webhooks	GET/POST/DELETE /webhooks/*	admin

WebSocket Streaming

Subscribe to real-time events, logs, and metrics:

{"type": "subscribe", "channel": "events"}
{"type": "subscribe", "channel": "logs", "process": "api-server"}

Outbound Webhooks

mhost api webhook add --url https://myapp.com/hook --events crash,restart --secret my-key

HMAC-SHA256 signed payloads with retry logic and dead letter logging.

Docker Integration

Manage Docker containers alongside native processes — one tool for everything.

mhost docker run nginx:latest --name web --port 8080
mhost docker list
mhost docker logs web
mhost docker stop web

All Commands

Process Management

Command	Description
`mhost start <app\|config>`	Start a process or ecosystem config
`mhost stop <app\|all>`	Graceful stop (SIGTERM -> grace period -> SIGKILL)
`mhost restart <app\|all>`	Restart with auto-recovery
`mhost delete <app\|all>`	Remove from process registry
`mhost list`	List all processes with status, CPU, memory, uptime
`mhost info <app>`	Detailed process info
`mhost env <app>`	Show environment variables
`mhost scale <app> <N>`	Scale to N instances
`mhost cluster <app> <N>`	Alias for scale
`mhost health <app>`	Show health check status per instance
`mhost config <app>`	Show process config as JSON
`mhost history <app>`	Show process event history

Groups

Command	Description
`mhost start --group <name>`	Start group in dependency order
`mhost stop --group <name>`	Stop group in reverse order

Logs

Command	Description
`mhost logs <app>`	Tail last 50 lines
`mhost logs <app> -n 200`	Tail last 200 lines
`mhost logs <app> --err`	Show stderr
`mhost logs <app> --grep "pattern"`	Filter by substring
`mhost logs <app> --search "query"`	FTS5 full-text search
`mhost logs <app> --where "level=error"`	Structured query
`mhost logs <app> --since 1h`	Time-range filter
`mhost logs <app> --format jsonl`	Export as JSON Lines
`mhost logs <app> --count-by level`	Aggregate by field

Notifications

Command	Description
`mhost notify setup`	Interactive channel setup wizard
`mhost notify list`	Show configured channels
`mhost notify test <channel>`	Send test notification
`mhost notify enable <channel>`	Enable a channel
`mhost notify disable <channel>`	Disable a channel
`mhost notify remove <channel>`	Remove a channel
`mhost notify events [channel]`	Show event types and subscriptions
`mhost notify start`	Start notifier as managed process

Metrics

Command	Description
`mhost metrics show <app>`	Current CPU, memory, uptime
`mhost metrics history <app>`	Time-series query
`mhost metrics start`	Start Prometheus /metrics exporter

Deploy

Command	Description
`mhost deploy <env>`	Deploy via git pull + hooks
`mhost rollback <env>`	Revert to previous successful deploy

AI

Command	Description
`mhost ai setup`	Interactive LLM provider setup (OpenAI/Claude)
`mhost ai diagnose <app>`	Analyze crash with root cause, fix steps, prevention
`mhost ai logs <app> "<question>"`	Natural language log search
`mhost ai optimize <app>`	Performance recommendations with config diff
`mhost ai config "<description>"`	Generate mhost.toml from plain English
`mhost ai postmortem <app>`	Generate incident report (Markdown)
`mhost ai watch`	Scan all processes for anomalies
`mhost ai ask "<question>"`	Ask anything about your processes
`mhost ai explain [config]`	Explain config in plain English
`mhost ai suggest`	Proactive improvement suggestions

Agent

Command	Description
`mhost agent setup`	Configure LLM provider, Telegram, autonomy
`mhost agent start`	Start the autonomous agent
`mhost agent stop`	Stop the agent
`mhost agent status`	Show agent config

Brain

Command	Description
`mhost brain status`	Fleet health scores with visual bars
`mhost brain history`	Past incidents and actions taken
`mhost brain playbooks`	Healing rules (built-in + auto-learned)
`mhost brain explain <process>`	Health score breakdown for a process

Cloud Fleet

Command	Description
`mhost cloud add <name> --host <ip>`	Add a remote server
`mhost cloud remove <name>`	Remove a server
`mhost cloud list`	List all fleet servers
`mhost cloud status`	Health check all servers
`mhost cloud deploy <server> <config>`	Deploy config to remote
`mhost cloud exec <server> "<cmd>"`	Run command on remote
`mhost cloud logs <server> <app>`	Stream remote logs
`mhost cloud restart <server\|all> <app>`	Restart across servers
`mhost cloud scale <server> <app> <N>`	Scale on remote
`mhost cloud sync <config>`	Push config to all servers
`mhost cloud ssh <server>`	Open SSH shell
`mhost cloud install <server>`	Install mhost on remote
`mhost cloud update <server\|all>`	Update mhost on remotes
`mhost cloud import <provider>`	Import from AWS/Azure/DO/Railway
`mhost cloud ai-setup "<desc>"`	AI infrastructure planning
`mhost cloud ai-diagnose <server>`	AI remote diagnosis
`mhost cloud ai-migrate <from> <to>`	AI migration planning

Cloud Native

Command	Description
`mhost cloud provision --provider <p> --name <n> --type <t>`	Provision a new cloud service
`mhost cloud services [--provider <p>]`	List services across providers
`mhost cloud service <name>`	Show service details
`mhost cloud cloud-deploy <name> --image <img>`	Deploy image to a service
`mhost cloud cloud-scale <name> <N>`	Scale a cloud-native service
`mhost cloud destroy <name> --provider <p> --confirm`	Destroy a service
`mhost cloud cost [--provider <p>]`	Show spending across providers
`mhost cloud drift [--fix]`	Detect configuration drift
`mhost cloud secrets set <svc> <key> <val>`	Set a service secret
`mhost cloud secrets list <svc>`	List service secrets
`mhost cloud secrets remove <svc> <key>`	Remove a service secret
`mhost cloud backup <service>`	Backup service data
`mhost cloud backup-list`	List all backups
`mhost cloud export <format>`	Export as Terraform/Compose/K8s

Bot

Command	Description
`mhost bot setup`	Interactive setup wizard (Telegram/Discord)
`mhost bot enable`	Start the bot (long-polling)
`mhost bot disable`	Stop the bot
`mhost bot status`	Show platform, enabled state, user lists
`mhost bot permissions`	Show role → user mapping
`mhost bot add-admin <id>`	Grant admin access
`mhost bot add-operator <id>`	Grant operator access
`mhost bot add-viewer <id>`	Grant viewer access
`mhost bot remove-user <id>`	Remove user from all roles
`mhost bot logs`	Show recent bot audit log

Production & Operations

Command	Description
`mhost reload <app>`	Zero-downtime reload (start new, health check, kill old)
`mhost dev <script>`	Development mode with file watching + auto-restart
`mhost dashboard [--port]`	Launch web dashboard (default port 9400)
`mhost bench <url>`	HTTP load test with concurrent workers
`mhost canary <app>`	Canary deployment — scale up, monitor, promote/rollback
`mhost snapshot create <name>`	Capture current process state
`mhost snapshot list`	List all saved snapshots
`mhost snapshot restore <name>`	Restore a snapshot
`mhost replay <process>`	Replay incident timeline from brain memory
`mhost link`	Show process dependency graph
`mhost cost`	Estimate cloud costs from process memory usage
`mhost certs [--url <urls>]`	Check SSL certificate expiry
`mhost sla <app> [--target 99.9]`	SLA uptime report
`mhost diff <env_a> <env_b>`	Compare two fleet environments/configs
`mhost share <app> [--port]`	Expose a local process via tunnel
`mhost run <file>`	Execute a recipe file (sequential mhost commands)
`mhost migrate --from <pm>`	Migrate from PM2 or other process managers
`mhost team`	Team management (coming soon)
`mhost playground`	Interactive tutorial

API

Command	Description
`mhost api start [--port]`	Start the API server
`mhost api stop`	Stop the API server
`mhost api status`	Show API server status
`mhost api token create`	Create API token
`mhost api token list`	List API tokens
`mhost api token revoke <id>`	Revoke a token
`mhost api webhook add`	Register a webhook
`mhost api webhook list`	List webhooks
`mhost api webhook remove <id>`	Remove a webhook
`mhost api webhook test <id>`	Test a webhook
`mhost api webhook failures`	Show failed deliveries

Docker

Command	Description
`mhost docker run <image>`	Run a container
`mhost docker list`	List mhost containers
`mhost docker stop/restart/logs/rm`	Container operations
`mhost docker pull <image>`	Pull image

Plugins

Command	Description
`mhost plugin install/list/remove/info`	Manage plugins

Templates

Command	Description
`mhost template list`	Show available templates
`mhost template init <name>`	Generate mhost.toml
`mhost init`	Auto-detect and generate config

Workspaces

Command	Description
`mhost workspace create/switch/list/current/delete`	Manage workspaces

Operations

Command	Description
`mhost audit`	Show audit trail
`mhost watch [config]`	Hot config reload
`mhost cron`	Cron schedule dashboard
`mhost limits <process>`	Resource limits
`mhost log-alert add/list/remove`	Log pattern alerts
`mhost hooks create/list/remove/test`	Incoming webhooks
`mhost status-page [--port]`	Public status page
`mhost rollback-process <name>`	Rollback config

Infrastructure

Command	Description
`mhost monit`	Launch TUI dashboard
`mhost proxy`	Show reverse proxy routes
`mhost ping`	Check if daemon is alive
`mhost kill`	Kill the daemon
`mhost save`	Save process list for resurrection
`mhost resurrect`	Restore saved processes
`mhost startup`	Generate OS boot script (launchd/systemd)
`mhost unstartup`	Remove boot script
`mhost self-update`	Update to latest version
`mhost completion <shell>`	Generate shell completions (bash/zsh/fish/powershell)

Architecture

┌──────────────────────────────────────────────────────────────┐
│                        mhost CLI                              │
│  start | stop | list | logs | monit | notify | ai | deploy   │
└──────────────┬───────────────────────────────────────────────┘
               │  JSON-RPC 2.0 over Unix Socket
┌──────────────▼───────────────────────────────────────────────┐
│                        mhostd (Daemon)                        │
│                                                               │
│  Process Supervisor     Health Checker     Reverse Proxy      │
│  - Spawn / signal       - HTTP / TCP       - Host routing     │
│  - Auto-restart          - Script probes   - Load balancing   │
│  - Backoff / circuit     - Status gate     - Auto-TLS         │
│  - Groups / deps                           - Sticky sessions  │
│                                                               │
│  Log Engine             Metrics            Notifications      │
│  - FTS5 search          - CPU / memory     - 8 channels       │
│  - JSON auto-parse      - Prometheus       - Throttle         │
│  - 4 external sinks     - Alerts           - Escalation       │
│  - Retention            - Auto-remediate   - Auto-resolve     │
│                                                               │
│  Deploy Engine          Scheduler          State Store        │
│  - Git pull / hooks     - Cron restarts    - SQLite           │
│  - Rollback             - Memory monitor   - Event history    │
│                                                               │
│  AI Intelligence        Cloud Fleet        Autonomous Agent   │
│  - OpenAI / Claude      - SSH fleet mgmt   - AWS / Azure      │
│  - Crash diagnosis      - Auto-import      - DO / Railway     │
│  - Config generation    - AI provisioning  - Remote deploy    │
│  - Optimization         - AI migration     - Fleet sync       │
│  Autonomous Agent                                             │
│  - Observe/Think/Act    - Telegram I/O     - Tool calling     │
│  - 3 autonomy levels    - Rate limiting    - Audit trail      │
└──────────────────────────────────────────────────────────────┘

Crate Structure (15 crates)

mhost/
├── mhost-core       Core types, state machine, protocol
├── mhost-config     TOML/YAML/JSON config parsing
├── mhost-ipc        JSON-RPC over Unix socket / named pipe
├── mhost-logs       Log capture, FTS5, rotation, sinks
├── mhost-health     HTTP/TCP/script health probes
├── mhost-notify     8 notification channels + throttle + escalation
├── mhost-metrics    Collector, time-series, Prometheus, alerts
├── mhost-proxy      Reverse proxy, TLS, ACME, load balancing
├── mhost-deploy     Git deploy, hooks, rollback, history
├── mhost-ai         LLM intelligence (OpenAI/Claude) — diagnose, optimize, ask
├── mhost-cloud      Remote fleet management — SSH, cloud providers, AI ops
├── mhost-bot        Telegram/Discord bot — role-based control, audit logging
├── mhost-tui        Terminal dashboard (ratatui)
├── mhost-daemon     Supervisor, handler, state store (mhostd binary)
└── mhost-cli        CLI interface (mhost binary)

Environment Variable Expansion

Use ${VAR} syntax anywhere in config files:

[process.api]
command = "${API_BINARY}"
env = { DATABASE_URL = "${DB_URL}", PORT = "${API_PORT}" }

[notifications.telegram]
bot_token = "${MHOST_TELEGRAM_TOKEN}"
chat_id = "${MHOST_TELEGRAM_CHAT}"

Cross-Platform

Feature	macOS	Linux	Windows
IPC	Unix socket	Unix socket	Named pipe
Signals	SIGTERM/SIGKILL	SIGTERM/SIGKILL	TerminateProcess
Startup	launchd	systemd	Task Scheduler
Memory	`ps`	`/proc`	WMI

Examples

The examples/ directory contains ready-to-run demo projects:

# Simple 3-process ecosystem
mhost start examples/mhost.toml

# Full-stack with health checks, groups, memory limits, cron
mhost start examples/full-stack.toml

Example	What it demonstrates
`node-api/`	HTTP server with `/health` endpoint
`express-api/`	REST API with CRUD, stats, health checks
`react-app/`	SPA with live dashboard UI
`python-worker/`	Background task processor with JSON logging
`bash-monitor/`	System metrics collector (CPU, memory, disk)
`cron-job/`	Periodic cleanup with report generation
`crasher/`	Unstable process for testing auto-restart
`flaky-api/`	API that degrades and recovers (5xx testing)

Community

Discord -- Join the mhost community for help, discussions, and feature requests
GitHub Issues -- Report bugs or request features
Contributing -- See CONTRIBUTING.md for how to contribute

License

MIT - Maher Al-Aqil

Name		Name	Last commit message	Last commit date
.github		.github
crates		crates
docs/superpowers		docs/superpowers
examples		examples
homebrew		homebrew
npm-package		npm-package
scripts		scripts
website		website
.gitignore		.gitignore
CHANGELOG.md		CHANGELOG.md
CLAUDE.md		CLAUDE.md
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
CONTRIBUTING.md		CONTRIBUTING.md
Cargo.lock		Cargo.lock
Cargo.toml		Cargo.toml
Dockerfile		Dockerfile
LAUNCH_POSTS.md		LAUNCH_POSTS.md
LICENSE		LICENSE
README.md		README.md
SECURITY.md		SECURITY.md
mhost.toml		mhost.toml
rust-toolchain.toml		rust-toolchain.toml

Uh oh!

Latest commit

History

Folders and files

Repository files navigation

🌐 mhostai.com · Pricing · Docs · Playground

Why mhost?

Installation

Quick Start

Start any process

See what's running

Manage processes

Persist across reboots

Ecosystem Config

Health Probes

Process Groups & Dependencies

Auto-Restart & Circuit Breaker

Notifications

Quick Setup (CLI)

Manage Channels

Supported Channels

Alert Events

Throttling & Escalation

Log Engine

Built-in full-text search (SQLite FTS5)

JSON auto-detection

External Log Sinks

Retention Policies

Metrics & Prometheus

CLI

Prometheus endpoint

Alert Rules & Auto-Remediation

Reverse Proxy

Deploy Engine

Cloud Fleet Management

Add Servers

Auto-Import from Cloud Providers

Remote Operations

AI-Powered Cloud Operations

Fleet Config (~/.mhost/fleet.json)

Cloud-Native Provider Integrations BETA

Supported Providers

Quick Start

Secret Management

Backups & Export

Chat Bot Control

Setup

Chat Commands

Permission System

CLI Commands

Security Features

Bot Config (~/.mhost/bot.json)

AI Intelligence

Setup

Crash Diagnosis

Natural Language Log Queries

Generate Config from Description

Performance Optimization

Incident Post-Mortem

Ask Anything

More AI Commands

AI Config

Autonomous Agent

Setup

How It Works

Conversation

Autonomy Levels

Safety

Self-Healing Brain

CLI

How It Works

Built-in Playbooks

Health Scores

Zero-Downtime Reload

Dev Mode

Web Dashboard

Load Testing

Canary Deployments

Snapshots

SSL Certificate Monitoring

Fleet Config (`~/.mhost/fleet.json`)

Cloud-Native Provider Integrations ^BETA

Bot Config (`~/.mhost/bot.json`)