Can I track LLM usage and token costs with this skill?

Yes, it includes built-in queries to analyze the $ai_generation event, allowing you to track input/output tokens, model latency, and total usage over time.

How does this skill handle sensitive PostHog API keys?

The skill uses a secure credential loading pattern that sources variables from a local .env file using a specific shell command, ensuring keys are never echoed in command history or output.

Does this skill support feature flag evaluation?

Absolutely. You can list all active feature flags or perform local evaluation to see exactly what flag state a specific user is experiencing.

Can I view my existing PostHog dashboards?

The skill allows you to list all project dashboards by ID and name, and provides a quick-access pattern to open specific dashboards in your browser.

PostHog Analytics & User Behavior

Name: PostHog Analytics & User Behavior
Author: dsifry

bydsifry

•

分析与监控

Queries PostHog production data to analyze user behavior, monitor feature flags, and track AI performance metrics.

The PostHog Analytics skill empowers Claude Code to securely bridge the gap between development and production data. By interacting directly with the PostHog API, Claude can investigate user journeys, identify friction points in onboarding funnels, and debug production issues using real-time event data. It includes specialized support for tracking LLM performance—such as token usage, latency, and costs—and provides a secure framework for managing feature flags and viewing project dashboards directly from the terminal.

主要功能

01Securely query production analytics using HogQL and PostHog APIs

02Analyze conversion funnels and user onboarding drop-off points

0329 GitHub stars

04Retrieve system error logs and AI fallback events for rapid debugging

05Monitor LLM performance including token usage, provider latency, and costs

06Evaluate real-time feature flag states for specific users or cohorts

使用场景

01Investigating the root cause of a specific user's reported issue using their event history

02Verifying feature flag rollouts and checking if a user has access to specific functionality

03Auditing AI generation costs and response times across different model providers

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dsifry/goodtogo posthog-analytics

For use in Claude.ai and ChatGPT