A workflow for choosing and running an A/B test on a Noibu-monitored online store. An A/B test compares two versions of a page or feature to see which performs better.

It is for finding test ideas, implementing a chosen variation, creating a code preview or Shopify preview theme, checking it, and starting the experiment.

It turns store data and site review into a selected experiment, while keeping people involved before the test is built or started.

ab-testing — Skill for Claude Code by Noibu/ai-plugin

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 Noibu/ai-plugin --skill ab-testing

Clone the repo

git clone --depth 1 https://github.com/Noibu/ai-plugin

Made for: Claude Code.

Or install noibu, the plugin that ships this one along with the rest of its 9 skills.

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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/noibu/ai-plugin/ab-testing/github.svg)](https://agentmods.dev/skills/noibu/ai-plugin/ab-testing)

Your own site

<a href="https://agentmods.dev/skills/noibu/ai-plugin/ab-testing"><img src="https://agentmods.dev/badge/skills/noibu/ai-plugin/ab-testing/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/noibu/ai-plugin/ab-testing"><img src="https://agentmods.dev/badge/skills/noibu/ai-plugin/ab-testing.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 99 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 17,556 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.

Model	Per session	Once invoked
Fable 5.1	$0.00099	$0.17556
Opus 5.5	$0.00040	$0.07022
Sonnet 5.5	$0.00020	$0.03511
Haiku 4.5	$0.00010	$0.01756

Measured 12d ago against content hash 11b90dc3e1dc, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

ab-testing 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 12d ago.

The scan reads SKILL.md. This mod also ships 25 executable files (scripts/build_page.py, scripts/walkthrough/add_to_cart.js, scripts/walkthrough/build.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

src/skills/ab-testing/SKILL.md · 353 lines

How it starts

The opening of the file, as written. The whole thing — 353 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Noibu A/B test (find, pick, build, run)

Works best in Cowork, linked to a computer with the Claude desktop app open and Claude in Chrome connected: Claude's built-in browser reads the store and shows mobile previews at phone size, and Claude in Chrome takes the desktop previews and screenshots at full desktop size. When one browser is missing or gets blocked, the other covers for it, and without either the run continues on Noibu data alone; see Setup below before anything else.

Takes a Noibu-monitored store from "what should we test?" as far as the user wants: a code preview, a pull request, a Shopify preview theme, or a running experiment. Six stages, three human gates:

Resolve the store domain to its Noibu UUID (one call; nothing else runs before the first card)
Find opportunities → user picks one (the three-option choose-a-test card is the first thing the user sees; "Look for new A/B test opportunities" runs the finder half of this skill, and "Implement an existing A/B test" opens a second card of the tests already found); then, once a test is picked, detect the environment (git host, commerce platform, how code reaches production) and ask setup questions only when needed 1b. Summarize the picked test inline and the user picks what to automate (the run mode)
Draft the test in Noibu, write the variation code (with Noibu's implementation code in it), open a pull request
Build or locate a preview and run automated QA
Hand off: the user confirms and merges the pull request (full mode), or picks the next step (partial modes)
Verify the code is live, then start the test

Never merge, publish, deploy or start a test before its gate. The only writes before the merge gate are a Noibu draft test, a branch + pull request, and an unpublished / non-production preview — and a partial mode creates only the subset its button promised. The finder half writes nothing in Noibu at all: its published page and its cart round trip (one item added and removed) are its only side effects.

"Pull request" (PR) below means the host's equivalent: GitHub/Bitbucket/Azure DevOps pull request, GitLab merge request (MR). The cards always say "pull request" and never name a host; once the run is talking about a specific one (the hand-off summary, the merge gate), use the host's own word.

Read the full file on GitHub · 353 lines

Files

What ships with it

47 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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.

12d ago First seen · 353 lines · 99 tokens per session scan A 11b90dc3e1dc

Subscribe to this mod's changes

ab-testing is a skill published in the GitHub repository Noibu/ai-plugin (5 stars, last pushed 13d ago), licensed MIT. It adds 99 tokens to every session and 17,556 once invoked, about $0.0004 per session on Opus 5.5. 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 2026-10-30.

E-commerce Marketing and SEO

ab-testing next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
ab-testingNoibu/ai-plugin	Skill	5	13d ago	MIT	A	99	Claude Code
amazon-reviews-api-skillbrowser-act/skills	Skill	6.3k	7d ago	MIT	A	124	Claude Code, Codex
amazon-competitor-analyzerbrowser-act/skills	Skill	6.3k	7d ago	MIT	A	48	Claude Code, Codex
asc-subscription-localizationrorkai/app-store-connect-cli-skills	Skill	1.1k	4d ago	MIT	A	60	Claude Code
external-purchasesrshankras/claude-code-apple-skills	Skill	881	2d ago	MIT	A	98	Claude Code
food-orderBitterbot-AI/bitterbot-desktop	Skill	2.5k	3d ago	MIT	A	53	Claude Code, Codex
logistics-opsericrisco/rsc-harness	Skill	579	yesterday	MIT	A	127	Claude Code, Codex