Testery for Claude Code

Your agent
says it's done.
Where's the proof?

Produce shareable test evidence directly from Claude.

Install once and every run comes back with video, logs, screenshots, and bug reports you can hand to anyone — so you can automatically fix the tests that broke and report the defects that matter. Bundles the Testery MCP Server plus ready-made slash commands.

  • Video
  • Logs
  • Screenshots
  • Shareable bug reports
$ /plugin marketplace add testery/claude-plugin
$ /plugin install testery@testery
$ /reload-plugins

Run these in Claude Code. Free to install, works with the tests you already have, and authenticates with your existing Testery account — no API keys to manage.

Recommended Actions

Once installed, try the failed-run autofix workflow:

$ /testery:autofix-failed-test-run
A failed Testery test run rendered inside Claude Code, with fix, file bug, console, video, and rerun actions on the failing test

actual Claude Code conversation with interactive Testery results

From red to green, proven on Testery

Ask in plain English. Claude Code runs your suite on Testery, catches the real failures, fixes them, and reruns just those tests until the build is provably green. Fast and flake-free.

claude  ·  testery plugin
~/code/checkout-web on main

Preview changes. Verify results. Share the proof.

The Testery plugin bundles the Testery MCP Server together with curated slash commands and agents tuned for test automation workflows. Preview what a change does before you adopt it, verify the results independently of the agent that wrote the code, and walk away with evidence you can share — all backed by your live Testery account, without wiring up a single tool yourself.

image

Auto-fix failed test runs, then prove it

Point Claude Code at a failed run and let the plugin do the legwork. It pulls the failing run from Testery, investigates each failure to pin down the root cause, applies a fix in your working copy where it can, and files a shareable bug report for anything that needs a human — then reruns just those tests and hands back the evidence that the build is green.

$ /testery:autofix-failed-test-run

Just describe what you want

Just ask in plain language — your AI assistant does the rest.

“The AI just refactored checkout. Run the suite on staging and show me what actually breaks.”
“Why did the last run fail? Group the failures and show me the video for each one.”
“Rerun only the failed tests from the most recent regression run and prove the build is green.”
“File a bug report for that checkout failure with the logs and screenshots attached.”

What's in the plugin

A complete toolkit for driving Testery from Claude Code.

  • Testery MCP Server, preconfiguredThe hosted Testery MCP Server is wired up automatically, giving Claude Code direct access to your projects, runs, results, and environments.
  • Failure triage workflowsAsk Claude Code to pull the latest results, group failures, and explain what went wrong — with the right context already loaded.
  • Bug reports filed from a failureWhen a failure needs a human, the plugin turns it into a bug report with the evidence attached instead of leaving you to reconstruct it.
  • One-command setupAdd the marketplace, install the plugin, reload. Authentication uses your existing Testery login.
  • Run & rerun commandsSlash commands to kick off a test run for any project and environment, or rerun just the failed tests from a previous run.
  • Shareable evidence from every runVideo, logs, and screenshots come back with the run, so the proof is something your team, your auditors, and your agent can all read.
  • Project & suite managementList, create, and update the projects, test suites, and environments configured in your Testery account.
  • Always up to dateThe plugin updates from the Testery marketplace, so new commands and capabilities show up automatically.
image

Don't take the agent's word for it

Your agent proposes; Testery verifies. The plugin runs your tests independently of the model that wrote the code, so a passing build is a fact you can check rather than a claim you have to trust — and every run leaves behind video, logs, and screenshots anyone on your team can read.

Set up in seconds, secure by default

Three commands add the plugin and connect it to your account. Authentication is OAuth-based and tied to your existing Testery login, so your whole team can get started securely without copying API keys around.

image

Who it's for

Developers

Verify what your agent just shipped without leaving Claude Code. Install once and drive Testery in natural language.

QA & test engineers

Investigate failures faster with workflows that pull results, group failures, and turn the real ones into bug reports with the evidence attached.

Platform & DevEx teams

Standardize how your org proves AI-written code works, by rolling out one shared, versioned plugin everyone drives from Claude Code.

Frequently asked questions

How is the plugin different from the MCP server?

The Testery MCP Server is the connection layer that lets an AI assistant call Testery. The plugin packages that server together with curated slash commands and agents for Claude Code, so everything is preconfigured the moment you install it.

How do I install it?

Run /plugin marketplace add testery/claude-plugin, then /plugin install testery@testery, then /reload-plugins in Claude Code, and authenticate with your Testery account.

What does "shareable test evidence" mean?

Every run on Testery records video, logs, and screenshots for each test. The plugin pulls that evidence back into Claude Code and can file it as a bug report, so proving what broke does not depend on anyone reproducing it by hand.

Do I need a separate Testery account?

You use your existing Testery account. The plugin authenticates with OAuth and respects your account's permissions.

Can I use Testery without the plugin?

Yes. If you prefer to wire things up manually or use a different MCP client, you can add the Testery MCP Server directly. The plugin is the fastest path for Claude Code users.