agentless_monitor 1.0.0

Monitor 1000+ servers without installing any agent. Uses SSH, SNMP, and REST API protocols to collect CPU, RAM, Disk, Network, Process and Service metrics. Auto-discovers new hosts in CIDR subnets. Ze [...]

agentless_monitor

Monitor 1000+ servers without installing any agent.
SSH · SNMP · REST API · Auto-Discovery · Real-time Alerts · Pure Dart

---

Why agentless_monitor?

Pain point	Traditional agent-based monitoring	agentless_monitor
Setup time	Install agent on each server (days/weeks)	Seconds — just add credentials
Agent overhead	50–200 MB RAM per host	Zero — reads /proc via SSH
Windows/network devices	Agent often incompatible	SNMP works everywhere
New VMs auto-discovered	Manual registration	✅ Automatic CIDR scanning
Dart/Flutter SDK	Not available	✅ First-class support

---

Installation

# pubspec.yaml
dependencies:
  agentless_monitor: ^1.0.0

dart pub get

---

Quick Start — 5 lines

import 'package:agentless_monitor/agentless_monitor.dart';

final monitor = AgentlessMonitor();

monitor.addTarget('192.168.1.0/24', TargetConfig(
  via: [Protocol.ssh, Protocol.snmp],
  interval: Duration(seconds: 30),
  discover: DiscoveryMode.auto,
  ssh: SshAuthConfig.keyFile(username: 'admin', keyPath: '~/.ssh/id_rsa'),
  snmp: SnmpConfig.v2c(community: 'public'),
));

monitor.metricsStream.listen((r) =>
  print('${r.host}: CPU=${r.cpu?.usagePercent.toStringAsFixed(1)}% '
        'RAM=${r.ram?.usagePercent.toStringAsFixed(1)}%'));

await monitor.start();

---

Feature Overview

Protocols

Protocol	Works with	Metrics
SSH	Linux, Unix, macOS	CPU (/proc/stat), RAM (/proc/meminfo), Disk (df), Network (/proc/net/dev), Processes, Services
SNMP v1/v2c/v3	Windows, Cisco, Juniper, VMware ESXi, any SNMP device	CPU, RAM, Disk, Interfaces, Processes, Custom OIDs
REST API	Prometheus node_exporter, Docker, Kubernetes, custom endpoints	Any JSON metric, Prometheus exposition format

Auto-Discovery

monitor.addTarget('10.0.0.0/16', TargetConfig(
  discover: DiscoveryMode.auto,   // Scans subnet every 5 min
  via: [Protocol.ssh, Protocol.snmp],
  ssh: SshAuthConfig.password(username: 'ops', password: 's3cr3t'),
));

monitor.discoveryStream.listen((d) =>
  print('🆕 New host: ${d.host} (${d.deviceType.name})'));

Alerts

// Use built-in presets (CPU > 80/95%, RAM > 80/90%, Disk > 80/90%, etc.)
monitor.useDefaultAlertRules();

// Or define your own
monitor.addAlertRule(AlertRule(
  id: 'db_cpu_hot',
  name: 'DB CPU > 70%',
  metric: MetricType.cpu,
  operator: CompareOperator.greaterThan,
  threshold: 70.0,
  severity: AlertSeverity.warning,
  forDuration: Duration(minutes: 5),   // Anti-flap hysteresis
  hostFilter: '10.0.1.*',             // Only DB servers
));

monitor.alertsStream.listen((alert) =>
  print('[${alert.severity.name}] ${alert.host}: ${alert.message}'));

Target Config Presets

// Linux server
TargetConfig.linux(ssh: SshAuthConfig.keyFile(username: 'admin', keyPath: '~/.ssh/id_rsa'))

// Windows Server
TargetConfig.windows(snmp: SnmpConfig.v2c(community: 'public'))

// Cisco / Juniper router
TargetConfig.networkDevice(snmp: SnmpConfig.v2c(community: 'private'))

// VMware ESXi
TargetConfig.esxi(snmp: SnmpConfig.v2c(community: 'public'))

// Cloud VM (Prometheus node_exporter)
TargetConfig.cloudVm(api: ApiAuthConfig.prometheus(baseUrl: 'http://10.0.0.5:9100'))

History & Aggregation

// Latest snapshot
final latest = monitor.getLatestResult('10.0.0.5');

// Last 1-hour trend
final history = monitor.getHistory('10.0.0.5', period: Duration(hours: 1));

// Average CPU over an hour
final agg = MetricAggregator();
final avgCpu = agg.avgCpu('10.0.0.5', period: Duration(hours: 1));
final peakCpu = agg.peakCpu('10.0.0.5');

---

Architecture

AgentlessMonitor (singleton engine)
│
├── TargetRegistry      — host registration + state tracking
├── MonitorScheduler    — per-host periodic timers
├── AutoDiscoveryEngine — CIDR scanning + port probing
│
├── SshCollector        — /proc/** reads via dartssh2
├── SnmpCollector       — GetBulk via dart_snmp
├── ApiCollector        — HTTP/REST via package:http
│
├── MetricAggregator    — circular in-memory time-series
├── AlertManager        — threshold evaluation + hysteresis
└── AlertHistory        — queryable event log

Reactive Streams

Stream	Type	Description
metricsStream	Stream<MonitorResult>	Every poll result
alertsStream	Stream<AlertEvent>	Alert fires / resolves
discoveryStream	Stream<DiscoveryResult>	New hosts found
statesStream	Stream<Map<String, TargetState>>	Full fleet state after each cycle

---

Configuration

AgentlessMonitor(config: MonitorConfig(
  maxConcurrency: 200,            // Parallel SSH/SNMP workers
  connectionTimeout: Duration(seconds: 10),
  globalRetryCount: 3,
  enableDiscovery: true,
  discoveryInterval: Duration(minutes: 5),
  persistMetrics: false,          // Set true for Hive persistence
  maxHistorySize: 1440,           // 24h at 1-min intervals
  logLevel: 2,                    // 0=none … 4=verbose
));

---

Running Tests

# All unit tests
dart test test/unit/

# Specific test file
dart test test/unit/cidr_parser_test.dart

# With coverage report
dart test --coverage=coverage/
dart pub global run coverage:format_coverage --lcov --in=coverage --out=coverage/lcov.info

---

Pricing Model (SaaS tier suggestions)

Plan	Max Hosts	Suggested Price
Free	25	$0
Starter	100	$50 / month
Pro	500	$200 / month
Enterprise	Unlimited	$0.50 / host / month

Formula: max(0, hosts − 25) × $0.50 / month

---

Comparison

Feature	agentless_monitor	Zabbix	Datadog	CheckMK
Agent required	❌ Never	✅ For most	✅ Required	✅ Required
Setup time	Minutes	Days	Hours	Days
SSH support	✅ Full	⚠️ Limited	✅	✅
SNMP v1/v2c/v3	✅	✅	⚠️	✅
Dart/Flutter SDK	✅	❌	❌	❌
Open source	✅ MIT	✅ GPL	❌	⚠️ Community
1000 hosts cost	~$487/mo	Free (self-host)	~$2000/mo	~$3000/mo

---

File Structure

agentless_monitor/
├── lib/
│   ├── agentless_monitor.dart        ← Single import barrel
│   └── src/
│       ├── core/                     ← Engine, config, result, scheduler
│       ├── targets/                  ← Registry, config, state, groups
│       ├── protocols/
│       │   ├── ssh/                  ← Connector, collector, commands, pool
│       │   ├── snmp/                 ← Connector, collector, OID lib, v3, traps
│       │   └── api/                  ← Connector, collector, auth, parser
│       ├── discovery/                ← Auto engine, scanner, prober, result
│       ├── metrics/                  ← CPU, RAM, Disk, Network, Process, Service
│       ├── alerts/                   ← Rule, manager, event, evaluator, history
│       ├── storage/                  ← MetricsStore, TimeSeries, Snapshots
│       └── utils/                    ← CIDR, intervals, retry, pool, logger
├── test/unit/                        ← 7 test files, ~100 test cases
├── example/bin/cli_example.dart      ← Terminal dashboard
├── example/lib/main.dart             ← Flutter widget example
├── CHANGELOG.md
├── LICENSE                           ← MIT
└── README.md

---

License

MIT © 2026 agentless_monitor contributors