Three seeds over the building authority's front door, then three more
over a variant where the two bare needs became doors of their own.
The diagnosis held: six of the eight front-door misroutes went to the
appeal, a concrete sentence standing among abstract categories, which
collected everyone with any sort of grievance whatever it was about.
Behind a door of its own it collected none.
The site got no better for it - 77.3% against 80.0%, inside the spread.
Two offers went from findable to hidden: the firm whose insurance
certificate had lapsed found "to be on the list, and then to file
without asking again" as a sentence on the front door, and fails every
seed once it is behind a door called "Work as a firm".
So the leak was not costing what it looked like it was costing, and the
rewrite is rejected on measurement. The mixed front door stays.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A municipal building authority is the first sector that the line a
case walks could not describe. An application is returned for the
drawing it was missing and walks the same stretch again; an appeal
ends the moment the authority changes its own decision, without ever
reaching the board; an applicant may withdraw from anywhere, while a
refusal may only come once the papers are in.
`reached_from: { <stage>: [<stage>, ...] }` says where a stage is
entered from when the line does not. A stage named there is entered
from exactly those stages and nowhere else, so saying where a refusal
may come from also says where it may not - which `turned_down_from`
could not, being one list for every bad ending at once.
Found by replaying fifteen real building cases through portal's own
engine against a throwaway JetStream: it refused three legal paths and
accepted two illegal ones. Portal needed no change - hand-writing the
correct graph into aggregates.yaml, cycle and all, was accepted and
enforced exactly. The rebuilt site now produces that same graph from
the plan, and the access policy follows it.
Also: a public view is of a bucket, not of a need. A permit register
holds the owner's small application and the approved firm's large one
and is one page either way; the rule refused to publish a bucket the
moment any role filed into it. A bucket only a role ever writes to is
still refused.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`type: location` on a field is portal 0.4.0's map pin, stored as
lat/lon/zoom. The reference tells the model it needs `optional: true`,
since a map needs a browser that can draw one.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Whatever reads the mailbox publishes on portal.mail.received; portal
appends the note after checking the record exists in a bucket this
site has and the sender is the address the record itself holds. A
note never moves a case: a mailbox is the one door anyone may knock
on, so it cannot drive a state machine even if someone forges their
way past the checks. The desks show the conversation beside the case.
Tested: the address parser, every refusal (a stranger, no address to
compare, an empty reply, the same reply twice, an unknown contract),
and the whole path through a real JetStream - the note lands, the
state does not move, and no event is written.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`type: location` puts a pin on a map and stores {lat, lon, zoom}; a
record that holds one renders as a still, on a desk and in any
resource card. `place:` on a question is the counterpart to `event:`:
event makes a page an occasion with a time, place makes it somewhere,
shows the map under the question, and opens any location field on the
page there instead of on the globe.
Every tile and every still is fetched by portal and served from this
origin (src/maps.rs, MapLibre vendored in public/), so a page with a
map on it still talks to nobody but us: no third-party request, no
consent banner, and the token never leaves the server. MAPBOX_TOKEN
unset turns the whole thing off, the way GARAGE_* does for uploads.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A page is read by the people who own it, and serde_json sorts keys, so
every file put `name` at the bottom under `action` and `consequence`.
With preserve_order the file reads the way it was built: name, then
description, then what the card does.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`requires:` on a need is one sentence naming what the person must
have or be for the alternative to work - a membership, a reference
number, the link from a mail. It renders as the card's first feature,
above the road, who-reads-it and the form, because a condition a
reader meets only after filling the form in has already cost them the
filling in. Whether the section exists is read from the plan, never
from its wording: under a lift every slot holds a token and a token is
never empty, which would conjure a section the live page lacks and
shift every feature after it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Five features were emitted empty, as layout: a door's strip, a desk's
list, an invited list, a public list, an invite form. Each now has a
title and a line that says what a row is and what the buttons under it
do, or what the list leaves out - true from the plan before a model
writes a word, and a slot the model can improve. Desk cards, their
links and the desk page itself get real fallbacks too. The lift no
longer copies an empty live string over a fallback, which would have
emptied every new slot on the next rebuild.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A stray placeholder or a lost link was only fixed when a model wrote
the mail, so a words file that already carried one kept it through
every later build. The pass now runs once on the finished
aggregates.yaml words, covering --words, --words-from and the model
alike; the football club's three mails lose the `{chain}` they were
shipped with.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The mail prompt said "short"; it now says how short. The football
club's enrolment mails halved in length and read easier, and the
vel's evened out, with every link kept.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The second Norwegian case: level with its earlier build on routing
inside the spread, desks 13 of 14 with outcomes in words, Norwegian
mails from Sonnet in one pass.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Live vel front door against a cold, bureaucratic twin: friendlier 9
of 10 to the live page, none to the twin, both orders. So the earlier
no-preference between the live doors and the rejected rewrite was a
verdict, not blindness; and judged clarity and measured routing are
different things.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Re-planning the contractor showed three wasted rounds: `moves_by: {
checking: submitter }` written as a bare word, a group without a
label, and the two featured owner roles left with nothing to play,
which the builder only warned about after the fact. The bare word and
the missing label now parse (a label defaults to the id), and a plan
that parses with spectator roles is sent back once with the finding;
on the contractor, round three fixed both.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A need may say which unfinished stages a case can still be turned
down from; the machine no longer offers the bad end from every stage.
uhhm's collaborators are declined before they start and depart after,
as the live site has it. The uhhm plan gets stages_text and three
desk buttons reworded: every uhhm desk scores full marks.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
`stages_text` on a need gives each stage plain words in the plan's
language. The fallback mails, the road line and the desk buttons use
them, and needs.yaml carries them as `stage_words` so the desk score
describes an outcome the way the site would. The vel's plan has them;
its desks score 26 of 26 where the id-based outcome scored 24.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The first desk score offered every button of the bucket, where portal
shows a row only the moves from its own state; that scored the live
vel 17 of 26 and sent a rewrite of the neighbour-help desk chasing an
instrument fault. With the row's buttons only: 24 of 26, and the
rewrite loses on the fair test too. Noted: an accent-stripped state id
(ikke_lost) reads as English even to a strong picker.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The inside of the business reads button labels and nothing else. For
every desk and bucket with two or more buttons, a worker is told what
just happened to a case, in the state's own words, and picks the
button; scored right of total with what each miss was taken for. The
vel's neighbour-help desk scores 2 of 4 and the notice board 1 of 2:
the first numbers a desk has ever had.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Four of the planned contractor's five needs had max_fields equal to
their required fields: the model counted the form and called it the
person's tolerance. `plan` now says so when most needs do, and the
reference tells the model to decide the number before writing the
form.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The one manual step left after `plan` was mapping the attached
personas' `need:` to whatever ids the model chose. A second call now
does it: the model sees the plan's needs and the persona texts and
returns ids only, after the plan is final, so no persona wording can
reach a page. An id the plan lacks is ignored and the reader is told.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
`does`, `shows`, `mail`, `mails`, `invites_by`, `events`,
`attended_by`, and the measured guidance on doors: the plan reference
the model reads lagged the builder by a week, so a planned site had no
opener, no mails and no onboarding.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The trainer had no CI. Every push now runs the tests and builds every
plan under plans/ against the iris release the live sites run, and
lints the onboarding site.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
iris instead of question_lint everywhere; the skill's procedure is
plan, build, score: the opener and the look-around, mails and the
sender, moves for the person, doors only when measured, the lift
through a rebuild, the language and link guards, the descriptions
score over the headings score, and the spread a change has to beat.
README documents --llm-for.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A model asked for Norwegian sometimes answers in English. The commonest
function words of each language are counted across a reply, and one
that reads as the other language is refused with a line saying so.
Tomter Vel's six mails, worded by Sonnet in Norwegian, are round 2 of
runs/tomtervel-opener-2026-09-23 and live.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Sonnet wrote the ten uhhm mails and replaced the decide link with a
bare {chain}. keep_links puts the fallback's link line back into any
mail that lost it. The mails read easier than the fallbacks (LIX 9 to
33 against 18 to 28) and say something.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The words model can be pointed at a subset of pages (`aggregates.yaml`
for the mails alone) while everything else keeps the words it has.
`rate --repo` rates each mail's prose on its own, LIX and length,
since a person reads a mail with no card around it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Every score so far was one run of a small visitor. `answer --seed n`
answers the same tasks with other shuffles and sampling, and
runs/spread.sh prints a run's score per seed, so a change can be told
from the visitor's mood.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The opener that says what the business does and the look-around card
lead nowhere: they are what the page says before its options, so a
task and a simulation step carry them as `page` context and offer
only the cards that lead somewhere. Five personas per round had been
"choosing" the opener and counted as misroutes.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
iris now hands a task the cards that lead nowhere as `page` context;
the answerer shows it before the options and the rater before the
cards, and the gutted twin loses it like everything else. The Tomter
Vel plan keeps its four doors and gains the opener, the mails and the
member's own move: flat scored worse and the doors stay.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The fallback mail phrases move into a `Phrases` table, the one place a
language other than English appears in this code; the `navn` special
case goes (a form names the person's field `name`). Test fixtures
speak English: traffic, board, hazards, report, fixed.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
`--words-from` matched page by page, so flattening Tomter Vel lost
every heading its door pages carried. Now a second pass pools every
alternative in the repo and hands a rendered card the pooled card
with the same strong identity (bucket, own page, invite group; never
a bare action like `/`). Two things stood in the way and are fixed:
needs.yaml is no longer a source of words (its `via` named the door),
and a live feature's `name` no longer blocks a positional match, so
placeholders and guidance lines come along too. Rebuilding the vel
flat from the live repo now lifts all ten headings and every hint.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
