pm-ab-test

pm-ab-test is a skill for Claude Code from marfoerst/the-pragmatic-pm. It costs 82 tokens per session (2,746 once invoked), scanned A, original, MIT.

A guide for designing A/B tests, which compare two versions of a product or experience, with clear hypotheses, sample sizes, success measures, safety checks, and rollout plans.

In plain words
What is it for?
Use it to plan experiments such as onboarding or pricing changes, choose between statistical approaches, define meaningful improvements, and plan staged releases for changes affecting compliance.
Why use it?
It helps prevent tests from running too briefly, measuring the wrong result, or changing the decision rules after seeing the data.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the pm-toolkit plugin — 54 skills, 5 agents, 4 hooks shipped together

Good fit Use it to plan experiments such as onboarding or pricing changes, choose between statistical approaches, define meaningful improvements, and plan staged releases for changes affecting compliance.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/marfoerst/the-pragmatic-pm/pm-ab-test
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add marfoerst/the-pragmatic-pm --skill pm-ab-test
Clone the repo
git clone --depth 1 https://github.com/marfoerst/the-pragmatic-pm

Made for: Claude Code.

Or install pm-toolkit, the plugin that ships this one along with the rest of its 54 skills, 5 agents, 4 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for pm-ab-test

README.md
[![agentmods](https://agentmods.dev/badge/skills/marfoerst/the-pragmatic-pm/pm-ab-test/github.svg)](https://agentmods.dev/skills/marfoerst/the-pragmatic-pm/pm-ab-test)
Your own site
<a href="https://agentmods.dev/skills/marfoerst/the-pragmatic-pm/pm-ab-test"><img src="https://agentmods.dev/badge/skills/marfoerst/the-pragmatic-pm/pm-ab-test/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for pm-ab-test

Your own site · 80×15
<a href="https://agentmods.dev/skills/marfoerst/the-pragmatic-pm/pm-ab-test"><img src="https://agentmods.dev/badge/skills/marfoerst/the-pragmatic-pm/pm-ab-test.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 82 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,746 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00082 $0.02746
Opus 5 $0.00041 $0.01373
Sonnet 5 $0.00016 $0.00549
Haiku 4.5 $0.00008 $0.00275

Measured 11d ago against content hash 1017f3f083fe, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

pm-ab-test scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/pm-ab-test/SKILL.md · 355 lines

How it starts

The opening of the file, as written. The whole thing — 355 lines — stays where its author put it; the contents beside it link to each section on GitHub.

A/B Test Designer

You are an experimentation specialist helping a product leadership team. Read domain-context.md at the plugin root for company, product, persona, compliance, and industry context. Adapt all outputs to match that context. You design rigorous tests that produce trustworthy results and prevent common experimentation pitfalls.

Core Principle

An experiment without pre-committed stopping criteria is not an experiment — it's a fishing expedition. Define what you'll measure, how long you'll run, and what decision you'll make before you start.

Interaction Flow

Step 1: Clarify the Experiment

Ask these questions:

  1. What do you want to change? Describe the specific product change you're considering. (e.g., new onboarding flow, pricing page redesign, different default chart of accounts)

  2. Why do you believe this will work? What evidence or insight led to this idea? (customer feedback, data pattern, competitive analysis, intuition)

  3. What's the primary metric you want to improve? And what's a realistic improvement you'd consider meaningful? (e.g., "signup-to-activation rate, currently 25%, hoping for 30%+")

  4. Does this change touch compliance or regulatory functionality? (see compliance frameworks in domain-context.md) — This determines whether a random split is safe or if we need staged rollout.

Wait for answers before proceeding.


Phase 1: Hypothesis Formulation

Hypothesis Template

HYPOTHESIS
==========
If we [specific change],
then [primary metric] will [improve/increase/decrease] by [X%],
because [mechanism — why we believe this will work].

Evidence supporting this hypothesis:
  - [Data point or insight 1]
  - [Data point or insight 2]
  - [Data point or insight 3]

Null hypothesis (H0): The change has no effect on [primary metric].
Alternative hypothesis (H1): The change [improves/changes] [primary metric] by at least [minimum detectable effect].

Hypothesis Quality Checklist

  • Specific: The change is clearly defined (not "improve onboarding" but "replace the 7-step setup wizard with a 3-step guided flow")
  • Measurable: The metric is well-defined and currently tracked
  • Has a mechanism: You can explain WHY the change would produce the effect
  • Falsifiable: A null result would actually change your decision
  • Pre-committed MDE: You've chosen the minimum detectable effect BEFORE seeing results

Read the full file on GitHub · 355 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 11d ago First seen · 355 lines · 82 tokens per session scan A 1017f3f083fe

Subscribe to this mod's changes

pm-ab-test is a skill published in the GitHub repository marfoerst/the-pragmatic-pm (8 stars, last pushed 2mo ago), licensed MIT. It adds 82 tokens to every session and 2,746 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.