Posthog
github.com/posthog/posthog- Category
- Developer Tools
- Rank
- No. 138Tools index
- Listed in
- #7 Monitor product and usage
- Type
- TOOL
- Builder
- posthog
- GitHub
- 39.7k stars
- Latest release
- desktop-v0.61.284
- Date
About
Comprehensive developer platform with analytics, feature flags, and AI-powered debugging assistant.
What it does
PostHog collects product events and session context, then lets teams inspect behavior, traffic, errors, logs, experiments, surveys, and AI traces together. Data arrives through a web snippet, SDKs, or an API. Teams can query it with visual tools or SQL, automate responses, and control features for selected users or cohorts.
Why it's ranked here
The appeal is consolidation with credible operational depth. Analytics can connect directly to recordings, errors, logs, warehouse data, experiments, and rollout controls. Agent access through MCP, editors, Slack, web, and desktop extends that context into diagnosis and proposed fixes. The trade is a broad, infrastructure-heavy system whose supported path clearly favors PostHog Cloud.
What's good
It supports both automatic event capture and deliberate instrumentation, serving quick adoption and precise measurement. Session replay adds direct behavioral evidence to analytics. Feature flags and experiments connect observation to controlled delivery. External data can be queried beside product data, while pipelines transform events and route them to other tools or webhooks. SDK coverage spans major web, mobile, and backend stacks.
Tradeoffs
The open-source hobby deployment targets roughly 100,000 events per month, carries no customer support or guarantees, and is not presented as the path for larger workloads. The repository also separates an enterprise directory under another license, so this is not uniformly permissive source. Its many product areas and substantial Python, Node, Go, Rust, Kafka, Redis, and database machinery imply meaningful operational complexity.
How to use it well
Use it when product and engineering teams want one evidence loop: instrument behavior, inspect funnels or SQL, replay failures, track errors, test changes, and release through flags. Start with Cloud unless hobby-scale self-hosting is an explicit requirement. Connect agents through MCP only after useful product context exists. It does not provide supported, guaranteed large-scale open-source hosting, and it does not replace your external warehouse when batch export remains necessary.
Technical notes+
README.md defines Cloud as the recommended deployment and a Linux Docker hobby installer as the self-hosted option. package.json identifies a pnpm monorepo requiring Node 24, with Turbo, Playwright, frontend tests, formatting, linting, schema generation, and product builds. pyproject.toml pins Python 3.13.13 and exposes a Django-heavy backend with Kafka, Redis, ClickHouse, Celery, data tooling, MCP, and multiple model-provider clients. nodejs/src/index.ts selects specialized ingestion, recording, error, log, metric, trace, and plugin servers by configured mode; nodejs/src/server.ts assembles CDP consumers and shared infrastructure. livestream/main.go implements an Echo HTTP service consuming Kafka, optionally distributing through Redis pub/sub, exporting Prometheus metrics, and shutting down gracefully. cli/src/main.rs and cli/src/lib.rs expose a Rust CLI with API proxy, source-map, debug-symbol, ProGuard, and dSYM modules. posthog/__init__.py initializes the Celery application and configures Snowflake state.
Observed
- License
- MIT expat for the repository except the ee directory, which has a separate license; a separate FOSS repository removes proprietary code and features.
- Packaging
- pnpm monorepo for JavaScript and TypeScript workspaces, plus a uv Python workspace and a Rust CLI.
- Install surfaces
- PostHog Cloud, a Linux Docker hobby deployment, a JavaScript web snippet, SDKs, and an API.
- Interfaces
- Web, desktop, Slack, MCP-compatible agents, SDKs, API, and Rust CLI.
- Platform coverage
- SDKs are documented for web frameworks, React Native, Android, iOS, Flutter, Python, Node, PHP, and Ruby.
- Runtime structure
- The supplied source shows Python and Node services, a Go livestream service, and a Rust command-line client.
Read from README.md, package.json, pyproject.toml, livestream/main.go, posthog/__init__.py, cli/src/lib.rs, cli/src/main.rs, nodejs/src/index.ts, nodejs/src/server.ts.
What it can do
Track and analyze user events automatically
Web application with PostHog integration → Event data and user behavior analytics
Create custom analytics dashboards with visualizations
Event data and user-defined metrics → Interactive charts and graphs showing product insights
Query analytics data with SQL
SQL queries and event data → Custom analysis results and data tables
Control feature rollouts with feature flags
Feature configurations and user segments → Targeted feature releases to specific users
Monitor web traffic and user sessions
Website with tracking code → Web analytics reports and session recordings
Debug application issues with AI assistance
Error logs and application context → AI-powered debugging suggestions and solutions
Manually instrument custom events
Custom event definitions and tracking code → Specific user interaction data and metrics
Intel on Posthog
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.