Guide · Agent skill
How to use the Replica Clone Tester agent skill (replica-test)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The replica-test skill is a set of instructions that an AI coding agent (such as Claude) loads so it can test an app clone you have built. It reads the user-flow map from replica/recon.md, writes a numbered test plan, generates Playwright end-to-end tests where automation is possible, and records reproduced bugs in a fixed report format. This guide is for developers who already have a clone and a local dev server, and it explains how to install the skill, trigger it, and use it safely.
Replica Clone TesterJakeschincariol/replica-skill · MIT Download the skill Quick start
- 1Download and unzip the skill. Download the replica-test ZIP from this page and unzip it into ~/.claude/skills/ or your project's .claude/skills/ folder.
- 2Check the prerequisites. Confirm you have Node.js and npm, a replica/recon.md file, and a local dev server running with seed data.
- 3Install the test tooling. In your project, run npm i -D @playwright/test, then npx playwright install chromium, and add @axe-core/playwright for accessibility scans.
- 4Ask for a test plan. Say something like "test my clone" so the agent writes replica/test-plan.md from the recon flows.
- 5Run the automated tests. Run npx playwright test to execute the generated specs against the local clone.
- 6Review bugs and fix S1 and S2 first. Read replica/bugs.md, check severities and evidence, and ask the agent to fix the serious bugs using the failing-test-first loop.
What is an agent skill, and what does replica-test make the agent do?
An agent skill is a folder with a SKILL.md file and supporting templates. When the skill is installed, the agent reads it and follows its steps when your request matches the skill's description.
replica-test works in five steps. First it writes a plan: for every flow in the recon map it lists a happy path, edge cases (such as empty input, double clicks, expired sessions and mobile width) and negative cases (such as a declined card). Second it automates what it can with Playwright, one spec per flow, run against your local dev server with seed data. Third it covers what cannot be automated, such as real emails or live payments, with a manual click-through. Fourth it writes bugs to replica/bugs.md with an ID, severity, steps, expected and actual results, and evidence. Fifth it runs a fix loop for the most serious bugs.
How do you install the replica-test skill?
Download the ZIP for replica-test from this page and unzip it. You can install it in one of two places.
For your user account, unzip it into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder. Use the project location if you want the skill to apply only to that codebase.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills (upload)
How do you trigger the skill with natural requests?
You do not need a special command. The skill is meant to trigger when you ask for testing in plain words. The source lists these phrases: "test my clone", "find bugs", "QA this", "click through everything", "write e2e tests" and "does it work". It also triggers after the /replica-build or /replica-backend steps.
Be specific about the flow or screen if you only want part of the app tested, for example "write e2e tests for the booking flow F01".
- "Test my clone"
- "Find bugs in the booking flow"
- "Write e2e tests for F01"
- "Does it work?"
What do you need before running it?
The skill depends on several things being in place. Missing any of them will stop the work.
Your project needs Node.js and npm, a Playwright Chromium browser, the @axe-core/playwright package for accessibility scans, and a local dev server with seed data for the clone. It also expects replica/recon.md from an earlier replica workflow step. A browser tool for the manual click-through is optional.
The source gives these install commands for the test tooling:
- Node.js and npm
- replica/recon.md from a prior replica workflow step
- A local dev server with seed data
- @axe-core/playwright for accessibility scans
- Optional: a browser tool for the manual step
npm i -D @playwright/test && npx playwright install chromium
npx playwright testWhat safety rules does the skill follow?
The skill runs scripts and uses the network, so review what it will run before you approve it. Its central rule is to test your clone, not the original app. It must never load test, fuzz, script or hammer the original app's servers. Using the original by hand as a normal user is allowed.
Only use this skill on an app you own. Do not point it at production sites that handle real payments or customer data. The tests use seed data and the Stripe test card 4000 0000 0000 0002, but running them against live systems is risky. Bugs are only reported when reproduced; unconfirmed suspicions go on a separate "to check" list.
- Test your own clone only, never the original app's servers
- Do not run against production with real payments or customer data
- Only reproduced bugs go in bugs.md
What are the output files and how do you read a bug report?
The skill writes replica/test-plan.md, replica/bugs.md and end-to-end specs in your project's e2e/ folder. Each bug has an ID, a severity from S1 (data loss, security hole, wrong payments or a blocked core flow) to S4 (cosmetic), the flow and case it belongs to, exact steps, expected and actual results, evidence, and a status such as open or fixed in a commit.
The skill ends with a summary of cases run, passed, failed, bugs by severity and fixes so far. It says to ship nothing with an open S1, and suggests /replica-diff as the next step.
- S1: data loss, security hole, payments wrong, core flow blocked
- S2: a feature broken with no workaround
- S3: broken with a workaround, or visibly wrong
- S4: cosmetic
What people use it for
- Writing Playwright end-to-end tests for a booking flow from an existing recon map
- Generating a numbered test plan with happy paths, edge cases and negative cases for each flow
- Checking every screen of a clone for console errors, server errors and accessibility problems
- Producing a severity-ranked bug list with steps and evidence before shipping a clone
- Building a manual checklist for emails, OAuth and payments that cannot be automated
Frequently asked questions
How do I install the replica-test agent skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for your user account, or into a project's .claude/skills/ folder. In the Claude apps you can also upload it under Settings → Capabilities → Skills.
What does replica-test require to run?
It requires Node.js and npm, the Playwright Chromium browser, @axe-core/playwright, a local dev server with seed data for your clone, and a replica/recon.md file from a prior replica workflow step. A browser tool for the manual click-through is optional.
Is replica-test free to use?
The source lists the skill's licence as MIT. Check the licence file in the repository for the full terms before using it in a commercial project.
Can I use replica-test to test someone else's website?
No. The skill says to test your own clone only and never to load test, fuzz, script or hammer the original app's servers. Using the original by hand as a normal user is fine.
What is the difference between an S1 and an S2 bug?
S1 covers data loss, security holes, wrong payments or a blocked core flow. S2 covers a feature that is broken with no workaround. The skill fixes S1 and S2 bugs first and does not ship with an open S1.
Does replica-test replace manual testing?
No. It automates what it can with Playwright and leaves items such as real emails, OAuth with real providers, end-to-end payments and visual glitches to a manual pass or checklist.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use replica-test if you have built an app clone with a recon flow map and a local dev server, and want a structured test plan, Playwright tests and reproducible bug reports before shipping. Do not use it on apps you do not own, on production systems with real payments or customer data, or on projects without a recon flow map, and choose a general-purpose testing workflow instead if you are not working from a replica.
Download the skill