Where Findy is
You are here: Decide the drawings (/design-advisor) It is the next thing to run.
1 feature(s) are planned and none has been through /build-pack yet. Nothing before it was skipped. One step ran ahead of its place: Decide the motion. The work stands; the steps before them still need doing.
15 features listed, 1 written up yet. 94 of 94 screens have their buttons and fields filled in. Click any station for what it does.
Done Not yet You are here Skipped: should have happened by now Ran ahead of its place Black: happens once Blue: repeats for every feature
1 Set up Once, ever
Create the folder (init.sh): Done. A setup script typed into the terminal. It makes the project folder and puts the first handful of commands in it. Run once, ever.
Create the folder
(init.sh)
Write the product spec (/spec-writer): Done, first ran 2026-08-23. Produces the spec: the document that wins every argument about what the product is, plus a short lookup version of it.
Write the product spec
(/spec-writer)
Set up the project (/bootstrap-project): Done, first ran 2026-08-26. Writes the instructions every session reads, the bookkeeping files, the design folder, the list of features, and installs the rest of the commands.
Set up the project
(/bootstrap-project)
2 Plan one feature For each feature, starting here
Lock the architecture (/pre-pack-architectural-surface): Done, first ran 2026-09-02. Walks one feature and everything it leans on, finds every architecture decision nobody has made yet, and puts them to you one at a time. The next step refuses to start without its receipt, on purpose.
Lock the architecture
(/pre-pack-architectural-surface)
Write the feature spec (/feature-plan): Done, first ran 2026-09-03. Writes four files for the feature: what it does, the shapes its data takes, every state each screen can be in (loading, empty, no signal, gone wrong), and how you know it passed. Screen designs need those states and cannot invent them.
Write the feature spec
(/feature-plan)
first feature only, then straight on
Choose the design tool (/design-platform): Done, first ran 2026-08-29. Records which tool draws the screens. Every design command reads that record. Once, and again only if you switch tools.
Choose the design tool
(/design-platform)
Load the design system (/design-system-push): Done, first ran 2026-08-29. Puts your colours, lettering and motion into the design tool before its first drawing. Skip it and the tool applies its own defaults, and you meet them as finished screens. Runs again whenever your colours or lettering change. Can run alongside the screen list; neither reads the other.
Load the design system
(/design-system-push)
List every screen (/screen-inventory): Done, first ran 2026-08-30. Lists every screen the whole app needs, once, from the product spec. The method allows that list to be made before the first feature is written up, so design is never left waiting; it is not allowed to be the only thing that runs. Then, as each feature is written up, its states and fields are added to the same list, and that feature's screens are drawn from it.
List every screen
(/screen-inventory)
Map the screen flow (/screen-inventory --flow): Done. Every screen as a grey block with the real words in it and an arrow off every tappable thing. The first moment the app can be looked at rather than read as a list, and the cheapest moment to move something.
Map the screen flow
(/screen-inventory --flow)
Draw the screens (in the design tool): Done. The design tool draws this feature's screens, from the screen list and the states in the feature spec. Not a command of this method. What it draws is STRUCTURE; the passes below decide how it all looks.
Draw the screens
(in the design tool)
one category at a time, all screens
Fill in each screen (/screen-inventory --detail): Done. Fills in what each screen actually contains: every button, field and tappable thing, and where each one leads. First of the category passes, and that order is not arbitrary: motion and sound attach to controls, so a screen that has not named its controls cannot have either of them decided.
Fill in each screen
(/screen-inventory --detail)
Decide the drawings (/design-advisor): You are here. Every illustration, character and icon, decided once for the whole app rather than screen by screen: what each one IS, before anything is made. /design-advisor has been run once, in August, and left no app-wide record of this, which is why the station is not ticked. The cast (the Finder, the Packer and the Keeper) is the part already drawn; the rest of the app's artwork is not.
YOU ARE HERE
Decide the drawings
(/design-advisor)
Decide the motion (/animate): Ran ahead. What moves, and how. Decided per layout family rather than per screen, because six layouts cover ninety-nine screens. IT COMES BEFORE THE ARTWORK IS MADE, and that order is a hard rule rather than a preference: what has to move decides what the artwork must contain, and a picture generated flat cannot be taken apart afterwards without inventing what was behind each piece. design/ANIMATION_RECIPES.md is the library's generic recipe book, not this app's decision, so it does not tick this station.
Decide the motion
(/animate)
RAN AHEAD
Make the artwork (/asset-forge): Not yet. Makes every drawing, icon and moving piece the two stations above have decided, in the right file type for where each one goes. Last of the making passes, because it needs both the what (the drawings pass) and the how it moves (the motion pass) settled first. assets/ is empty today.
Make the artwork
(/asset-forge)
Decide the sound (no command in the method): No command exists. What the app sounds like: the click when something is captured, the note when a thing is found, and silence everywhere else. THE METHOD HAS NO COMMAND FOR THIS ANYWHERE. That is a real hole, not a step you have finished, and it is drawn as a hole so nobody reads the empty space as done. It has to be walked by hand, or a command has to be written for it.
Decide the sound
(no command in the method)
NO COMMAND EXISTS
Check the designs (/design-check): Not yet. Reads the designs against the screen list and writes one report. That report saying PASS is the single thing that lets anybody write a line of code. A report is already on disk from 3 September, and it says in its own words that it is NOT a gate run, so the gate is still owed. This station is judged on the PASS line at the start of a line, never on a file existing and never on the words appearing mid-sentence.
Check the designs
(/design-check)
3 Build it The same feature
Plan the build phases (/build-pack): Not yet. Turns the feature into numbered build phases and stops. It writes no code. Several features can be planned and built side by side once their groundwork is done.
Plan the build phases
(/build-pack)
Preflight the design (/design-preflight): Not yet. Screens the phase against the design system and lists what artwork it needs. Visual phases only; needs the build plan first. Anything it finds missing is topped up with /asset-forge, which is the same command as the making pass in stage 2 and is why it has no station of its own here.
Preflight the design
(/design-preflight)
Build the code (/build-team): Not yet. A guide and a builder work through the phase. Code and tests.
Build the code
(/build-team)
Try it on the phone (/feel-check): Not yet. You hold the phone and say whether it feels right. Between visual phases.
Try it on the phone
(/feel-check)
4 Check and close Eight checks, then close the session
Code review (/pr-review-toolkit:review-pr): Not yet. Check 1 of 8. Reads the code.
Code review
(/pr-review-toolkit:review-pr)
Review the diff (/review-diff): Not yet. Check 2 of 8. Reads the diff, which is the exact set of lines that changed.
Review the diff
(/review-diff)
Security review (/security-review): Not yet. Check 3 of 8. Reads the change for security holes.
Security review
(/security-review)
Audit the diff (/diff-audit): Not yet. Check 4 of 8. Reads the change against what was meant.
Audit the diff
(/diff-audit)
App safety audit (app-safety-audit.py): Not yet. Check 5 of 8. Reads the finished app for what could go wrong.
App safety audit
(app-safety-audit.py)
Test the tests (test-strength.py): Not yet. Check 6 of 8. Breaks the code on purpose and checks the tests notice.
Test the tests
(test-strength.py)
Find temporary code (temp-code.py): Not yet. Check 7 of 8. Refuses while any temporary shortcut is undeclared or overdue.
Find temporary code
(temp-code.py)
Visual QA on a phone (/visual-qa, plus a phone test): Not yet. Check 8 of 8, and the only one that opens the running app. The other seven all come back clean on a build that launches to a blank screen.
Visual QA on a phone
(/visual-qa, plus a phone test)
Close the session (/close): Runs every session, first ran 2026-08-24. Files what was learned, updates the bookmarks, saves and uploads. Then back to stage 2 with the next feature. Since September 2026 it asks the guide for the next step rather than deciding itself.
Close the session
(/close)
5 Launch Once every feature is done
both at the same time
Pass the security gate (/security-floor): Not yet. The security floor for the app itself. Blocking before any real user.
Pass the security gate
(/security-floor)
App store preflight (/store-preflight): Not yet. App-store readiness for Apple and Google. Blocking before submission.
App store preflight
(/store-preflight)
After stage 4 the route goes back to stage 2 with the next feature, until every feature is done; then stage 5.
Any time, off the route
(/implementation-guide) (/secure-machine) (/setup-tools) (/guardian) (/diary)
How this page was made. Every mark was worked out from the project's own files on 5 September 2026: the stage from the session-start check, which commands ran from the session records, and steps that are not commands from the one file each of them leaves behind. Nothing is remembered. If this page and a bookmark file disagree, believe this one.