Owlmetry Documentation
Self-hosted metrics tracking platform for mobile apps. SDKs for Swift and Node.js, CLI for agents, and a REST API.
Owlmetry is a self-hosted observability platform built for the agentic development era. It captures events, metrics, and funnels from your apps and backend services, stores everything in a single PostgreSQL instance, and exposes it all through a REST API, CLI, and optional web dashboard.
Your agent can set up the entire integration, query events, diagnose issues, and act on what it finds -- without a human opening a dashboard.
Key Features
- Event ingestion with batch processing, deduplication, and gzip compression
- Structured metrics with lifecycle tracking (start/complete/fail) and aggregation queries
- Funnel analytics to measure conversion and drop-off across multi-step flows
- Issue tracker that clusters error events into actionable issues with regression detection
- Event attachments for uploading logs, screenshots, and crash dumps alongside error events
- A/B experiments with client-side variant assignment and server-side analysis
- Third-party integrations like RevenueCat that sync subscription data into user properties
- Agent-native API where every operation is available through
owl_agent_keys - Single PostgreSQL backend with monthly-partitioned event tables -- no Kafka, no ClickHouse, no Redis
Automate everything
The web dashboard is great for exploring data, but you shouldn't need to open it for routine work. Owlmetry exposes every operation -- creating projects, defining metrics, querying funnels, managing integrations -- through both an MCP endpoint and a CLI. For AI coding agents, we recommend the MCP endpoint -- configure it once and your agent always has the latest tools. Your agent can set up an entire analytics pipeline, investigate production issues, and act on the results without a human ever touching the dashboard.
| MCP | CLI | |
|---|---|---|
| Setup | One-time URL + key in client config | npm install -g @owlmetry/cli + login flow |
| Updates | Automatic — always matches the server | Manual npm update |
| Auth | Static Bearer token in headers | Login flow with email verification |
| Best for | AI coding agents (Claude Code, Cursor, etc.) | Humans, scripts, CI/CD |
| Requires | MCP-compatible client | Node.js 20+ |
Learn more: MCP setup · CLI installation
Documentation Sections
- Getting Started -- Set up your first project and start sending events in minutes. Connect your agent via MCP or CLI, then add an SDK.
- Concepts -- Core building blocks: events, metrics, funnels, experiments, projects and apps, authentication, and data mode.
- Swift SDK -- Native SDK for iOS, iPadOS, and macOS. Zero external dependencies. Covers events, screen tracking, metrics, funnels, identity, and experiments.
- Node.js SDK -- Server-side SDK for backends and serverless functions. Zero runtime dependencies. Covers events, metrics, funnels, identity, and serverless support.
- MCP -- Connect AI agents directly to Owlmetry via the Model Context Protocol. No installation, automatic updates, 52 tools covering every operation.
- CLI -- Command-line interface for agents and humans. Query events, manage resources, and inspect audit logs. Supports
--format jsonand--format table. - API Reference -- Complete REST API covering authentication, ingestion, querying, project/app management, metrics, funnels, issues, attachments, integrations, teams, and identity.
- Self-Hosting -- Deploy on your own infrastructure. Covers system requirements, PostgreSQL, nginx, SSL, pm2, and maintenance.