Skip to main content
skill CX Codex

debugging-experiments — Claude Code skill: setup and features

debugging-experiments — a Claude Code skill for investigating PostHog experiment results. It checks uneven or missing exposures against the experiment’s data and known causes.

no license 150 installs ★ 101 updated 2026-10-24

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add PostHog/ai-plugin --skill debugging-experiments -a claude-code
Open on GitHub

What debugging-experiments does

The skill investigates support questions about PostHog A/B test results, including uneven exposures, a variant with no traffic, bias warnings, and numbers that differ from SQL. It reads the experiment data, identifies a likely cause, then drafts an explanation with a fix and supporting figures.

For deeper checks, it loads the diagnosing-experiment-health library. It matches the proposed fix to the experiment’s state, whether draft, running, or stopped. This skill is not for creating experiments, configuring rollout or metrics, lifecycle questions, or debugging a misbehaving feature flag.

What it helps with

  • trace an uneven exposure split to its cause
  • find out why a variant receives no traffic
  • investigate missing or unusually low exposure counts
  • compare PostHog experiment results with SQL
  • explain a bias warning and outline a fix

Example prompts for the agent

  • Investigate why the variants in this experiment have different exposure counts.
  • One variant has no traffic. Check the data and explain what is happening.
  • Compare this PostHog experiment with my SQL results and find the discrepancy.

How to install debugging-experiments

1Skills CLI (Vercel)

npx skills add PostHog/ai-plugin --skill debugging-experiments -a claude-code

2Manually: put the files in place

~/.claude/skills/debugging-experiments/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/debugging-experiments/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about debugging-experiments

How do I install debugging-experiments in Claude Code?

Run npx skills add PostHog/ai-plugin --skill debugging-experiments -a claude-code. The skill file is at ~/.claude/skills/debugging-experiments/SKILL.md.

What does debugging-experiments do?

It reads PostHog experiment data to investigate result issues, finds a likely cause, and prepares an explanation with a fix and figures that support it.

What kinds of questions trigger debugging-experiments?

Questions about experiment results, such as uneven exposures, a variant receiving no traffic, or a mismatch with SQL. It is not intended for experiment creation or rollout configuration.

Similar skills

All skills for Code review and refactoring →
systematic-debugging Jesse Vincent systematic-debugging is a Claude Code and Codex skill that blocks fixes until root cause analysis is complete. It activates on test failures, bugs, or unexpected behavior. CX Codex +1 306,118installs debugging-and-error-recovery addyosmani The debugging-and-error-recovery Claude Code skill traces failures to their root cause and checks the fix. It activates for failing tests, broken builds, unexpected errors, or behavior that no longer matches expectations. CX Codex +1 79,600installs brainstorm-experiments-new phuryn brainstorm-experiments-new turns a new product idea into an XYZ hypothesis plus 2-3 pretotypes, each with a metric and a success threshold. Use it before writing code, to see whether anyone wants the thing. CX Codex +1 3,100installs brainstorm-experiments-existing phuryn The brainstorm-experiments-existing skill helps plan experiments to validate assumptions about an existing product before committing to full builds. Use it to test hypotheses with prototypes, A/B tests, spikes, and other lean methods. CX Codex +1 3,000installs debugging-strategies Seth Hobson The debugging-strategies skill gives an agent a step-by-step way to investigate bugs, crashes, and performance problems. Use it when a project behaves unexpectedly. CX Codex +1 13,414installs parallel-debugging Seth Hobson The parallel-debugging skill investigates complex bugs by testing competing hypotheses in parallel. Use it when an issue has several plausible causes or crosses multiple components. CX Codex +1 10,000installs

From the same set82

The whole set →
modeling-product-usage-metrics modeling-product-usage-metrics builds retention, frequency, and lifecycle models using PostHog HogQL views or dbt data marts with built-in tests. CX Codex +1 175installs modeling-dimension-tables modeling-dimension-tables builds star-schema dimension tables in PostHog HogQL views or dbt projects so you avoid repetitive joins across your metrics. CX Codex +1 174installs modeling-warehouse-foundations modeling-warehouse-foundations sets core guidelines for building PostHog data models via native HogQL views or an external dbt setup. CX Codex +1 172installs instrument-metrics The instrument-metrics Claude Code skill adds PostHog metric capture to a codebase. It runs after feature work or PR review, and when a request calls for counters or other metrics. CX Codex +1 154installs context-layer-health-check The context-layer-health-check Claude Code skill checks a context wiki after changes are consolidated. It triages missed entries, conflicting claims, and outdated priorities. CX Codex +1 150installs understanding-billing-usage understanding-billing-usage is a Claude Code skill that investigates PostHog usage and spend through Billing MCP tools. It handles questions about spikes, changes, and the products or projects behind usage. CX Codex +1 150installs building-html-canvases The building-html-canvases Claude Code skill builds PostHog canvases with semantic HTML, CSS, and browser APIs. It runs when a canvas request is routed to a plain HTML implementation. CX Codex +1 149installs composing-grid-canvases The composing-grid-canvases Claude Code skill arranges PostHog grids from reusable widgets. It activates for requests to add, resize, move, or remove a placement, or assemble a whole canvas. CX Codex +1 149installs

Project sponsors

Companies that keep this analytics open