Every app featkpr has read, and what it found in them.

1. 1 · the app

   BookStack

   open-source wiki. Its code read, its screens crawled, its features named.

2. 2 · a feature, in Activity

   View audit log of system activity

   2 screens found · 4 tests written, 4 proven by the wrong-user check.

3. 3 · part of 2 goals, linked by route
   • Open the audit log (index)proven · complex · /settings/audit, a flow ends here
   • Open the favourite (index)no test yet · moderate · /favourites, a flow ends here
4. 4 · 2 flows end on it
   • 1. /
     2. /favourites

     “View the list of items I've marked as favourites.”

     this flow · waiting on a person to keep or drop it
   • 1. /
     2. /settings
     3. /settings/{category}
     4. /settings/audit

     “Starting from the homepage, navigate through settings to view the audit log.”

     this flow · waiting on a person to keep or drop it

On the Flows board this feature is one goal of its flows; in the goal records it is part of 2. The two counts of goals differ for this reason, and both are true.

Ask for a private demoFollow any feature like this on the call, on BookStack.

1. Settings that describe the app #

   How to check it out, start it on a throwaway copy and seed one user per role. Since 30 Sep you add the app yourself, as one flow: paste the address, give one token from a page that opens already filled in, then First read, Answers and Go. The screens below are the first design of 28 Sep; the flow as released differs.

   written for each app · a few files

   

   

   

   Starting the app on a throwaway copy is not in the flow yet, so it is still a folder we write per app: targets/bookstack/ product.toml ae/checkout.sh ae/start.sh ae/routes.sh owners.toml glossary.toml

   Shlink's folder has product.toml, owners and a glossary, and no start script yet: that is why its column stops at two steps.

2. Read the code #

   The app's own route table, each route with the file and line it comes from. This is the only part that depends on the stack.

   written for each app · no model call

   GET /books/{bookSlug}/permissions
     module   Permissions
     handler  PermissionsController@showForBook
     quote    Route::get('/books/{bookSlug}/permissions',
                [PermissionsController::class, 'showForBook']);
     where    routes/web.php, line 74, commit 0f5164ec

   343 of 359 routes read for BookStack; the audit names the 16 it missed.

3. Above the line, written for each app.Below it, the same code for every app: a new app adds no new stop.

4. Name the features #

   A model groups routes into features a person would recognise, module by module. Most of the model spend is here; drafting goals later costs cents.

   the same for every app · $0.99 for BookStack, 22 Sep

   

   298 features in 13 modules (20 Oct 2026).

5. Crawl, then draft flows #

   A throwaway copy is walked, signed in as the admin today, its create forms are sent, and each goal gets the flow a person takes to it, screen by screen.

   the same for every app · model calls only to draft the goals

   

   167 screens as Admin (25 Sep 2026) · 149 flows ending at 89 features (20 Oct 2026).

6. Test, check, and count what was missed #

   Tests are written per route and role and run. Each is sent again as the wrong user, who should get the opposite answer, and must then fail. Then featkpr starts again with no help, and each step is compared with a key (for routes, the list the reader itself uses).

   the same for every app · no model call for the run

   

   276 of 311 proven by the wrong-user check (20 Oct 2026, 03:29 UTC) · 605 of 2,372 items missed (26 Sep 2026).

Read from production 20 Oct 2026: which reader read each library app, and the routes and screens it read there at the app's newest commit. A cell is filled when an app read with that reader reached that step. A reader in the code that no library app was read with is listed as built, never as read. The stacks with no reader are the common ones our research met while counting apps (below); asking for one is how it gets a plan item.

1. featkpr runs its own goals on a test copy of itselfshipped 30 Sep: A disposable featkpr on the homelab, seeded with made-up data, signed in as the owner and as nobody, with a script per goal rendered from its command. featkpr's own goals now get runs and wrong-user flips, away from production.
2. Featkpr's own features named in production: 146, with 45 goalsshipped 30 Sep: part of "featkpr maps its own code: its parts, and the goals a change can alter"
3. Node and Django readerslater, no week yet: Django, Hono, React Router and Next.js readers shipped on 30 Sep and 1 Oct; Uptime Kuma and Paperless-ngx are not read yet

The order and the weeks are on the roadmap, with the long-term idea of a public library of every open-source app's features on its ideas board. Weeks ahead are estimates and move.