Betterstack

The REST API at uptime.betterstack.com/api/v2 follows conventional REST patterns: monitors, monitor_groups, heartbeats, incidents, on_call_calendars, and status_pages are top-level resources. CRUD operations use standard HTTP verbs with JSON payloads. Pagination uses cursor-based pagination with next and prev URLs in the response. The Logs API at logs.betterstack.com uses a separate endpoint for log ingestion and querying. Status page updates support incident creation and component status changes. Heartbeat monitoring (cron job verification) uses simple HTTP pings to unique URLs — agents just hit the URL on success. The API surface is small enough to learn quickly but lacks the depth of enterprise monitoring APIs.

Better Stack (formed from Better Uptime and Logtail) provides three integrated products: uptime monitoring (HTTP, TCP, DNS, cron checks), incident management with on-call scheduling, and log management with live tail and querying. For agents managing service health, the appeal is the integration — a single platform handles check → alert → escalate → log investigation. The REST API covers monitor CRUD, incident management, on-call calendar, and status page updates. Log ingestion supports HTTP, syslog, and various collector agents. The platform is simpler than Datadog or New Relic but covers the 80% case for monitoring-focused agents. Free tier includes 10 monitors and basic logging.

API authentication uses Bearer tokens generated in the dashboard. Tokens grant full API access within the team — no permission scoping or read-only tokens. Log ingestion uses separate source tokens, one per log source, which provides useful isolation (a compromised log token can't access the monitoring API). No OAuth, no temporary credentials. Team members have role-based access (admin, member) for console access but API tokens aren't role-scoped. For agents, the practical implication is simple: one API token for monitoring operations, one source token per log source. The simplicity is adequate for single-team deployments and limiting for multi-tenant agent architectures.

Documentation is clean, modern, and well-organized for the platform's scope. API reference includes curl examples and response samples. Getting started guides are step-by-step with screenshots. The documentation accurately reflects the product's scope — it doesn't try to position Better Stack as a full observability platform, which sets appropriate expectations. Integration guides cover common log sources (Vercel, Heroku, Docker, various languages). SDKs are available for Ruby, Python, Node, Go, and Elixir (Logtail clients). Community is smaller than enterprise monitoring platforms — fewer third-party resources. The product's simplicity means less documentation is needed, and what exists is generally sufficient. Dashboard UX is clean and fast.

API errors return JSON with errors[] array containing descriptive messages. Standard HTTP status codes: 401 for auth failures, 422 for validation errors, 429 for rate limits. Rate limits are generous (documented at 100 requests/minute for most endpoints). Monitor check intervals range from 30 seconds to 5 minutes depending on plan. Incident creation is automatic when monitors fail — agents don't need to poll for status changes; they can rely on webhook notifications. The platform's multi-region check infrastructure means monitoring reliability is decoupled from a single data center. Log ingestion accepts data optimistically — malformed logs are accepted but may not parse correctly. Alert notification channels include Slack, PagerDuty, email, SMS, and webhooks.