Pineapple on pizza

No data available

Feature ownership
Members

Roadmap

Recently shipped

Trace Gemini Interactions calls

AI observability captures foreground Gemini Interactions calls, including streamed responses, tool activity, usage, and supported media output. Agent and background interactions are still excluded because their completion lifecycle works differently.

Goals

Q4 2026 objectives

Goal 1: Improve SDK observability and diagnostics (Driver: Manoel Aranda Neto)

  • Benchmark SDK performance
  • Build dashboards to monitor SDK health and performance
  • Capture diagnostic events to investigate SDK issues
  • Add remote configuration support for SDK diagnostics
  • Enable or disable diagnostics globally, per organization, or per project

Goal 2: Expand SDK coverage and automate development with shared specifications (Driver: Dustin Byrne)

  • Define canonical SDKs for frontend, backend, and mobile
  • Run agentic loops for bi-directional sync of SDKs and specs, verified by a shared acceptance-test harness
  • Expand first-class SDKs and integrations:
    • Release @posthog/browser v2 GA
    • Release @posthog/next v1 GA
    • Release posthog-kmp v1
    • Release posthog-android-surveys-compose v1
    • Release the iOS SDK v4
    • Add Laravel support
    • Add Electron support
    • Add Angular support
    • Add Flutter desktop support through a confirmed community donation
      • Review the donated implementation
      • Test Flutter desktop support
      • Release Flutter desktop support

Goal 3: Bring mobile session replay to feature parity with web (Driver: Ioannis Josephides)

  • Decide whether to deprecate wireframe mode for mobile session replay, and proceed if agreed
  • Add v2 session replay triggers to mobile SDKs

Goal 4: Research how to capture product usage across agent-driven interactions (Driver: Anna Garcia)

  • Interview customers building agents, bots, MCP servers, and edge applications to understand what they need to measure, how they capture events today, and where our SDKs fall short
  • Evaluate capture options, including deriving analytics events from tool calls, capturing agent-driven interactions through WebMCP, and offering lightweight SDKs for edge runtimes and unsupported ecosystems
  • Share findings with the MCP and web teams, and recommend opportunities to test with customers
    • Pair with the Product Analytics team to explore their ideas and test them together where client libraries support is needed

Handbook

What the Client Libraries team does

  • SDKs - the knowledge to level these up are in our team

How to work with the Client Libraries team

Reach out in our Slack channel #team-client-libraries

Feature ownership

You can find out more about the features we own