Checks
Route: /checks (sidebar: Checks)
Is my app OK, and which check is red? Every automatic check Mushi Mushi runs on this app, in one grid, failing checks first. A sentence at the top names the checks that need you, with one button to open the first.
Each card has a dot and a state:
| Dot | State | Meaning |
|---|---|---|
| Red | Failed | The last run failed or could not finish. |
| Amber | N findings | The check ran and found something to fix. |
| Green | Pass | A finished run was read and found nothing. Shown with when it last ran. |
| Grey | Not run yet, Not set up, Status unavailable | Never a pass. Not set up comes with a Set up link. |
An Upgrade pill means the check needs a higher plan. AI prompts and CI runners are Labs: their cards appear once you switch them on under Settings → Display & sidebar → Labs. While you are on Checks or one of its pages, the sidebar lists every check under Checks.
The checks
App checks
Is the app itself sound: scheduled tests, user stories, audits, its blueprint, design tokens and content, a first-time user test, plus Mushi Mushi’s own health.
| Card | Route | What it covers |
|---|---|---|
| Scheduled tests | /qa-coverage | Browser tests of your user stories that run on a schedule and alert when one breaks. |
| Mushi Mushi health | /health | Is Mushi Mushi itself working: its AI calls and background jobs. Also your code size, schema changes and unusual spikes. |
| CI runners | /ci | Where your tests run, how long they wait for a machine, and what slows your fixes down. |
| Full-stack audit | /fullstack-audit | An end-to-end audit across the frontend, the API and the database. |
| User stories | /inventory | The things users can do in your app, found by crawling it, and which ones broke. |
| App blueprint | /recipe | What this app is built from — database, design system, routes, CI, deploy, env and integrations — and what changed. |
| Design system | /design | Design tokens, contrast pairs, type scale and off-token code. Token and rule edits open a draft pull request. |
| Content checks | /content | AI-generated and user-submitted content flagged as wrong or low quality. |
| First-time user test | /first-user-test | An AI that has never seen your app tries to sign up and use it; where a new user gets stuck and how long it takes. See UX runs. |
AI quality
How good the AI’s triage is: grading, lessons learned and the prompts behind it.
| Card | Route | What it covers |
|---|---|---|
| Triage grading | /judge | A second AI grades how well each bug was triaged and flags where it disagrees with the first. |
| Lessons | /lessons | Patterns learned from past bugs and fixes. |
| AI prompts | /prompt-lab | Edit and compare the prompts that turn bug reports into pull requests. |