# Turn your release checklist into tests that run themselves

> The checklist in the wiki gets skipped on busy days. Move each line into a plain-language test, run them from CI or a coding agent, and keep the evidence per release.

By Jason Nguyen, Founder, Testsy. Published 2026-11-21. https://testsy.dev/blog/turn-a-release-checklist-into-saved-tests/

## A checklist is not a control

Most teams have one: a page in the wiki titled "Before release". It lists the things someone should click through. On a quiet day it gets done. On a busy day, the day you most need it, it gets skipped. A guide on post-deploy verification says it in one line: [a checklist in a wiki is not a control](https://us.fitgap.com/stack-guides/automate-post-deploy-verification-so-releases-stop-relying-on-tribal-knowledge); a smoke suite that runs the same way every time is.

## Each line becomes a test

Take the checklist as it is. Most lines are already almost a test:

| Checklist line | Test |
| --- | --- |
| Sign-up works | Sign up with this run's email address, confirm from the email, and check the dashboard greets the new user |
| Can create a task | Create a task "Release smoke sample", reload, and check it is still listed with that title |
| Export is OK | Export the project report and check the downloaded CSV has the columns id, name and status |
| Billing page OK | Open Billing as the stored account "admin" and check the current plan and next payment date are shown |

The difference is the check: each one names what must be visible. That is what makes it repeatable, and what lets a result say exactly what it saw.

## Let a coding agent draft them

If your team uses Claude Code, the [release-check walkthrough](https://testsy.dev/docs/claude-code-release-checks/) shows the flow: give the agent one journey, ask it to draft the plan with `plan_test`, review the plan, then save and run it. The plan is shown before anything runs, so you check that it tests your requirement and not an easier one.

![The plan the assistant drafts: numbered steps, the checks, and the options.](https://testsy.dev/docs/img/plan-light.webp)

## Run them every release

Put the tests in one project. Then:

- **From CI**, after each deploy: [GitHub Actions](https://testsy.dev/blog/run-browser-tests-after-every-deploy-with-github-actions/), [GitLab](https://testsy.dev/blog/run-plain-language-tests-in-gitlab-ci/), [any CI](https://testsy.dev/blog/run-browser-tests-from-any-ci-with-one-file/).
- **From the app**, with **Run all** on the project.
- **From an agent**, with `run_project`.

## Keep the evidence

Each run records the commit and job it came from, and each result keeps its checks, values, replay and video for 365 days. When someone asks "did we check export before Tuesday's release?", the answer is a link, not a memory.

## What to keep manual

Exploratory testing, new features nobody has written expectations for yet, and anything that needs taste. The checklist's job was never those; it was the repeatable lines, and those are what automation should take.

This post is part of [Release testing for product teams](https://testsy.dev/blog/release-testing-for-product-teams/).
