Meticulous Test Skill: What It Does and How to Install It

Makes your agent build your frontend, upload it to Meticulous, run a visual regression test and review the changes.

Installs: 8.7K; on skills.sh
Stars: 10; on GitHub
Updated: Oct 2026; last repo commit

meticulous-test makes your agent run a Meticulous test run after a frontend change. It finds the build step in your CI config, builds the frontend, uploads the build, triggers a run against a base commit, and then hands the result to meticulous-review to classify each visual change as intended or unintended.

The same uploaded build can be re-run against different bases without rebuilding, and it works with uncommitted changes. It needs the meticulous command-line tool, a signed-in Meticulous project and a git repo. It begins by running meticulous-cli-update, which can install or update software.

Good for

Developers having an agent build a feature end to end before a PR
Teams already using Meticulous for visual regression testing
Reviewers who want each visual change marked intended or unintended

Install meticulous-test

Claude Code

Copy this prompt.

Install the meticulous-test skill in Claude Code. Run this command:

/plugin marketplace add alwaysmeticulous/skills
/plugin install meticulous@meticulous

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

meticulous-test calls other Meticulous skills, so install the whole set (the repo has 11). Full setup: app.meticulous.ai/docs/agents/setup. Plugin skills are named /meticulous:<skill-name>. After the plugin install, type /mcp and choose Authenticate for the Meticulous server. Choose one route; the skills CLI command targets Claude Code, Codex and Cursor together.

From the project README. Source.

What's in the skill

When it loads

Run a Meticulous test run after implementing a frontend change, then hand off to the `meticulous-review` skill to classify each visual change as intended or unintended. Uploads the build once — the same build can be re-triggered against different bases without rebuilding — and it works with uncommitted changes. Use when implementing a feature autonomously end-to-end before creating a PR.

Sections in SKILL.md

Step 1 -- Build the frontend
Step 2 -- Upload the build
Step 3 -- Trigger a test run
Step 4 -- Review the visual changes

Key facts

Author: Meticulous (the vendor's own repo, alwaysmeticulous/skills)
License: ISC (LICENSE, README and plugin.json all say ISC)
Installs on skills.sh: 8,674 (Nov 2, 2026)
GitHub stars (repo): 10 (Nov 2, 2026)
Last commit (repo): Oct 29, 2026
What installs: 1 file: SKILL.md. The repo holds 11 skills in all; this one calls several of them
Needs: The meticulous CLI (npm package @alwaysmeticulous/cli), a Meticulous account and project, and a git repo
Also needs these skills: meticulous-cli-update, meticulous-review
Works in: Claude Code, Codex and Cursor (README); the plugin also connects a hosted Meticulous MCP server

What it includes

Runs the CLI and skills update check unless it already ran in this conversation.
Installs or updates software; see the safety section.
Runs code · meticulous-cli-update (first)
Finds the build step in your CI config and runs the same build.
It checks .github/workflows, .gitlab-ci.yml or bitbucket-pipelines.yml for the upload-assets or upload-container step.
Runs code · Step 1: build the frontend
Uploads the build and prints a deploymentId; it does not start a run.
Use --appDirectory for build output or --localImageTag for a Docker image. A dirty working tree is captured as an ephemeral commit; untracked files are rejected.
Runs code · meticulous agent upload-build
Triggers a run for the deployment against a base and waits for it to finish.
The base is inferred from the repo or set with --baseSha. The final status is Failure when visual differences are found, which SKILL.md calls a normal completed verdict. --dontWaitForTestRunToComplete returns early.
Runs code · meticulous agent trigger-test-run
Hands the testRunId to meticulous-review to classify each visual change.
With uncommitted changes, always pass the explicit --testRunId.
Network · Step 4: meticulous-review

From the official docs

Permissions

Medium risk

This skill runs your build and uploads the result to Meticulous. Only use it on code you are allowed to send, and check the build commands first.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It uploads your build to Meticulous
upload-build sends your built frontend assets or Docker image to Meticulous. With uncommitted changes it records an ephemeral commit for them. Only run it on code you are allowed to send. github.com
It runs your project build commands
Step 1 tells the agent to build the frontend the way your CI config does. Check the build steps in that config before approving them. github.com
It installs or updates software for you
Every Meticulous skill starts with meticulous-cli-update, which can run npm install --global @alwaysmeticulous/cli@latest, npx skills add alwaysmeticulous/skills with -y, and plugin updates. SKILL.md says to recommend the command instead if you decline a permission prompt. github.com
The review step sends a feedback note to Meticulous
meticulous-test hands off to meticulous-review, whose last step runs meticulous agent submit-feedback with a short message, an outcome and the test run ID. Tell your agent to skip it if you prefer. github.com
The plugin route connects a hosted MCP server
The plugin bundles an .mcp.json for https://app.meticulous.ai/api/mcp and no hooks or agents appear in the repo tree. You sign in from /mcp in Claude Code. github.com

Technical details

Talks to	app.meticulous.ai
Sign-in and keys	Browser sign-in with meticulous auth login stores a token; METICULOUS_API_TOKEN or ~/.meticulous/config.json also work.
Scopes	Runs the meticulous CLI on your machine Uses your Meticulous sign-in or API token

Questions

How do I install meticulous-test?

Use the install box at the top of this page. The README installs the whole set of Meticulous skills, because this skill calls others. You also need the meticulous command-line tool and a sign-in, which the README gives as npm install --global @alwaysmeticulous/cli@latest then meticulous auth login.

Does meticulous-test work in the Claude app?

Not as a Claude app skill. It runs the meticulous command-line tool and works with your git repo, so use Claude Code, Codex or Cursor, which the README names. The plugin route also connects a hosted Meticulous MCP server.

Do I need a Meticulous account?

Yes. The skills use the Meticulous CLI or MCP server, and sign-in happens in your browser through meticulous auth login. An API token set in METICULOUS_API_TOKEN also works, according to the skills' update instructions. What Meticulous itself costs is not stated in the repo.

Is meticulous-test safe to use?

It runs commands and talks to Meticulous, so read what your agent proposes. It runs your project's build commands and uploads the build to Meticulous. Every Meticulous skill also starts with an update check that can install packages, and the review step it hands off to ends with a feedback note sent to Meticulous. See the safety section for details.

Is it official?

Yes. It is in Meticulous's own repo (alwaysmeticulous/skills), and the plugin manifest names Meticulous as the owner.

What if my CI is not on GitHub?

SKILL.md names the CI config to check for each host: .github/workflows/*.yml for GitHub, .gitlab-ci.yml for GitLab and bitbucket-pipelines.yml for Bitbucket. It reuses the build steps it finds there.

Information

updated Nov 2, 2026

Repository: alwaysmeticulous/skills
Folder: skills/meticulous-test

alwaysmeticulous/skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 6

alwaysmeticulous/skills README · github.com
Meticulous agent setup docs · app.meticulous.ai
meticulous-test SKILL.md · github.com
meticulous-cli-update SKILL.md · github.com
marketplace.json · github.com
plugin.json · github.com