The classification rides the API for free (taskView embeds task.Task),
so this is the read side only: a parked row shows what kind of
intervention it needs -- needs you, transient, terminal -- as a chip
with the answer in its tooltip, instead of a bare Parked every time.
Non-parked rows render nothing.
A stage filter is a command scoped to the attempt it was made on: the
rail set it against that attempt's stage list. selectAttempt and
performRetry left filters untouched, so a stage selected on attempt 2
carried into attempt 1 -- a same-named stage there rendered as selected
without ever being commanded, or the log stayed narrowed to a stage the
new attempt does not have. The stage filter now resets when the selected
attempt changes; the level filter is attempt-agnostic and survives.
The master-detail PRD absorbs the adversarial review's verified findings:
the reload claim is narrowed to the URL contract (filters are not in the
URL), the unimplemented scroll and pinned claims are dropped, the
unnamed-stage and level-filter contracts are stated, the rail inventory
is corrected, and the acceptance list gains criteria for the
command-not-binding rule, attempt-switch reset, and keyboard selection.
The first cut grouped consecutive same-kind events, but the real noise
shape is an interleave: one background cycle emits curator_run,
session-memory and curator_action events carrying the same payload, and
kind-keyed grouping left them all visible. The grouping key is now the
rendered detail plus the task, so one cycle of noise collapses to one
row regardless of which internal kind emitted each event, while events
carrying different information never merge. Verified against a live
daemon: 50 background events collapse to one expandable row.
The five tabs gave the stage rail less than a third of the screen and
disconnected the log stream from the stage that produced it. The stage
rail is now the page's master pane, always visible; the four context
panels are the inspector and own the tab bar. Selecting a stage commands
the inspector to the log tab with that stage's filter set; clicking the
selected stage clears it. Selection is a command, not a binding -- the
operator can change the filter afterwards.
No endpoint, RPC or contract changed. tab=stages still resolves, to the
log tab; every inspector tab and attempt deep-link survives a reload.
Below the shell's 900px frame breakpoint the rail stacks above the
inspector. Design of record: docs/prds/task-run-master-detail.md (Draft,
per-run panel contracts unchanged from task-run-detail.md).
The feed's 50 raw events were dominated by consecutive same-shape
background rows (session-memory cycles, runner calls), which buried the
milestones the dashboard exists to surface. Consecutive events sharing
one label and one task now collapse into a group row with an xN badge;
expanding it reveals its events in place. The grouping is consecutive,
not global, so interleaved kinds keep their order and a milestone is
never merged with an event that merely shares its kind.
The projection is a pure view over the store's raw stream (activity-group.ts,
table-tested); the store keeps every event. Group rows keep ActivityRow's
link contract, and the disclosure button is stop-propagated inside the
first cell so it never fights the row-level link.
The info token doubled as badge text on info-soft and as the log-level
label, and failed 4.5:1 there (4.29:1 dark over card, 4.45:1 over glass;
light theme 4.11:1 on canvas, 3.58:1 on its soft). Dark --info moves to
#B08BF8, light --info to #6F47A6, both clearing 4.5:1 on every surface
they land on. Light --fg-subtle moves to #675E6F for the wash-composited
canvas where the old value measured 4.44:1. Measured pairs recorded at
the token definitions.
Two halves of one boundary: a credential resolves to an identity, and an action
records which identity acted and whose authority it used.
resource server. internal/infrastructure/oidc verifies a provider-issued bearer
token against the provider's published keys, using an OSS OIDC library: archie
compares no signature, validates no claim and rotates no key itself, holds no
client secret and issues no session, so a compromised archie cannot mint an
identity. Audience is required rather than optional -- a token minted for another
service must not authenticate here -- and discovery happens at construction, so an
unreachable provider is a startup failure rather than a request-time surprise.
identity.Authenticate is the single place a credential becomes an identity, and it
refuses in four distinguishable ways: no credential presented, a credential that
did not verify (wrong key, expired, wrong issuer, wrong audience, malformed), a
verified subject with no identity bound to it, and an identity that may not act.
The order is deliberate: verification happens before any field of the token is
used, so nothing an unverified token carries can select an identity. A suspended
identity is refused even though its credential verified, because suspension is
archie's decision and must not depend on the provider withdrawing a token.
The binding is the second bridge beside the legacy-name alias: identity_subjects
keys (issuer, subject) so the same subject string from another issuer is a
different caller, and rebinding moves a subject so two identities can never claim
one provider subject. The contract Authenticate needs is SubjectResolver, narrower
than Repository, so a consumer that only authenticates does not take the whole
identity CRUD surface -- and staterpc.Client, which asserts Repository, does not
need a new RPC to satisfy it.
attribution. taskactions now records two identities per action: the acting
identity and the authorising principal, both on the event row as new columns added
through the existing migration table. The rule that the bug turned on is enforced
by the event kind describing the actor: a person's approval is human_approved, an
agent's is agent_approved, and an action with no verified actor is task_approved,
a task-level event that credits nobody. An agent acting under a person's standing
approval records itself as the actor and that person as the principal, so
"approved by came from me" is answerable from the record rather than asserted.
Scope and actor are separated: scope decides which task a caller may touch and is
never promoted into an actor, because the dashboard's credential is a shared token
and possession of it is not evidence that a person was at the keyboard.
That last point is a deliberate fidelity trade, and the reason the transport is
not wired: the dashboard action path passes no actor yet, so dashboard approvals
are now recorded as task_approved rather than claimed as a human's. Recording
nothing false is the requirement; naming the human verifiably needs the dashboard
to authenticate a token and resolve it, which needs two new state-store RPCs
(ResolveIdentitySubject, BindIdentitySubject) -- the next piece, not a rewrite:
the verifier is a one-method interface, so introspection would be a second
implementation of it.
Verified: task check exit 0; focused runs recorded in the handoff. Opaque
authelia_at_ tokens were inspected rather than assumed and cannot be verified
against public keys; infra is configuring access_token_signed_response_alg RS256
on the archie client, with audience https://archie.catlow.cloud and issuer
https://auth.catlow.cloud.
The header's secure-context note no longer reads as a dead end: 127.0.0.1 is
already a secure context, and mkcert gives a LAN address or hostname a
locally-trusted certificate, so a plain-HTTP origin is a configuration gap with
a known local fix. Production terminates TLS in front of the dashboard.
A WebMCP tool call cannot be attributed to an agent: the page can invoke its own
tools with executeTool, so an agent's call and a page script's call reach the
server identically. A durable record can honestly say only that the actor was
the signed-in identity and the source was webmcp -- a hint, not proof -- and
must never claim a human approved. Real agent attribution needs the agent to
authenticate with its own credential, which is the A2A work where an agent is
an Identity.
The module header now records the two ways the tools go silently missing: a
secure context is required, so a plain-HTTP dashboard exposes nothing even with
the origin-trial flag, and archie-ui embeds ui/dist, so rebuilding the bundle is
not deploying it and a stale process reports the tools as absent. It also
records that executeTool flattens a thrown execute error into a generic
UnknownError, which is why the confirm guard is pinned by a unit test rather
than by a browser run.
The three recovery tools now carry a required `confirm: true`, rejected before
the API is touched when absent. It is a deliberate-action signal, not a
security control -- an agent can pass it trivially, and the module header says
so. Its value is making the model state intent rather than reaching for a
mutation mid-sentence.
The module headers also record two things this work learned the hard way: the
API's three string/object shape traps (inputSchema and executeTool's input and
result are JSON strings, not objects), and the advisory status of the
annotations -- Chrome 152's getTools() reports only readOnlyHint and
untrustedContentHint and drops consequentialHint, and a page-invoked
executeTool gates nothing. Nothing here is enforced by the browser. The tools
run in the signed-in operator's session and cannot exceed that operator's
rights; that is the security model.
Register the dashboard's own tools with document.modelContext so an in-page
agent can read tasks, events, and daemon health, and recover a stuck task:
list_tasks, get_task, recent_events and daemon_health (read-only), plus
retry_task, stop_task and cancel_task (consequential). The human review gate
(approve/reject) and the destructive verbs (abandon/archive) are deliberately
not exposed.
Every tool calls the same ui/src/lib/api.ts client the interface calls, so the
agent and the UI exercise one set of endpoints, one CSRF header and one error
shape. Registration is framework-agnostic (lib/webmcp.ts machinery,
lib/webmcp-tools.ts list, lib/archie-webmcp.ts the api binding) and no-ops when
document.modelContext is absent, which it is outside the WebMCP origin trial.
Tools returning agent, operator or inbound content carry untrustedContentHint;
only the recovery actions carry consequentialHint.
Verified in system Chrome 152 with --enable-features=WebMCP: seven tools listed
with schemas and annotations, all four read-only tools executed against the live
dashboard, and the flagless run mounts the dashboard with no exception.
An unauthenticated document request answered with a plain-text 401 -- and
index.html, the bundle and every asset sit behind requireToken -- so a
bookmarked or bare URL was a dead end: the dashboard's own authentication
banner could never load to explain it. Document requests now get a
self-contained 401 page whose one-field form reuses the ?t= exchange, so a
pasted token sets the cookie and the redirect lands on the dashboard. A
rejected token re-renders the page with the reason instead of answering a
document with text.
API and event-stream callers keep the plain-text 401: classifyActionError maps
it to "session-expired", and an EventSource error body is never seen.
The auth banner gains a Retry that re-runs the reads which report their own
401s, so a recovered session clears it without a document reload. The retry is
extracted to lib/session-retry.ts with an in-flight guard and unit tests.
Inspector, Bindings and Mappings were three sibling destinations in a
dropdown, for what is one chain read in order: an event, the mapping that
names its fields, the binding that decides what runs. They are now three
tabs on /events, and the bar carries Events as a link rather than a menu
with one child sharing its name.
The navigation entry is always live and the tabs prune themselves: each is
a separate capability, so a composition that can back only some of them
still reaches the ones it can, and an unreported section stays visible.
Which tabs exist, and which one is showing, lives in tab-selection.ts and
is unit-tested without a browser; the page owns the components and the
address bar. The old paths survive as bookmarks that redirect to their tab,
so a link or a learned URL still lands somewhere that works.
Three pages become tab panels: they lose their page-level headers -- the
page names the page -- and keep their own actions in a right-aligned row.
CapturesHeader.vue is absorbed into the Inspector panel, which is where its
stream status belongs.
The control-plane descriptors now derive a JSON Schema from the Go type of
the document they describe (6ba10c0b), and nothing consumed it: the editor
rendered dashed keys, guessed no format, and synthesised a new row from a
literal skeleton per array path.
A pure reader (settings/resource-schema.ts) resolves the schema node for an
editor path -- nested keys, array items, and a keyed collection's
additionalProperties -- and the editor uses it for labels (title ahead of
the dashed-key fallback), a duration placeholder, a hint when the schema
carries prose, and the shape of a new row. A descriptor with no schema, or
one describing nothing, renders exactly as before.
Deletes ControlPlaneResourceCard.vue: 121 lines, zero references, the
abandoned schema-driven card and the only other consumer of schema_json.
The settings page now reads apply status beside each resource: every known
process with the version it applied, its error if it rejected the edit, and
whether it is still reporting at all. It re-reads after each save, so the page
reflects what the processes did with an edit rather than what was stored.
The server derives liveness only, because it knows whether a record went stale
but not which version the page is showing. The client compares against the
resource it is rendering, which is what turns a live record on an older
version into a pending restart.
A process that stopped re-stamping and one that never reported are separate
rows with separate wording, and neither is an error: the setting may be fine
and the process simply absent, which an operator acts on differently. A failed
apply-status read leaves the previous rows rather than taking the settings
page down with it.
state-store-contract.md gains the pair in its inventory and its count moves
from 45 to 47, with the note that this read is a list rather than the
config-snapshot pair's found=false singleton.
Closes archie-core-pskb.
2311529d deleted the overlay but not its documentation of record, leaving
docs/architecture/ asserting a subsystem that no longer exists.
safe-change-and-recovery.md was the dangerous one. It listed three ways back
from a bad runtime setting, all of them gone, so its recovery section was a
dead end. The replacement records the posture honestly: the control plane
fails closed, a stored value that will not validate stops the daemon starting,
and there is no escape hatch until archie-core-6zw0 lands.
configuration.md's ledger rows for the overlay store and the two dashboard
write routes become one row for the control-plane command. migration-decisions
marks the config-write descope superseded rather than deleting it, per
docs/prds/RULES.md.
Also removes leftovers the gate could not catch: a dashboard banner bound to
ReloadStatus.overlay_unavailable, a field the server stopped sending, and a
delivery-snapshot scan pointed at ApplyOverlayValues in a composition root that
no longer calls it.
The reload trace behind this found archie-core-ju85: SIGHUP republishes the
file document alone, reverting every database-owned setting until restart.
Recorded in the ledger as current behaviour.
resource_history has carried a row per change since the control plane landed
(value, version, actor, source, request ID, both versions, timestamp), but
nothing could read it. The audit trail existed and was invisible, and the PRD's
"shows failed changes and can correct or roll them back" had nothing to stand
on.
Restoring is deliberately not a command of its own. A revision carries the
value it held, so putting one back is an ordinary replace against the version
currently stored: the restore lands in the same history as every other edit,
under the same optimistic concurrency, instead of rewriting the past.
The revision value crosses the HTTP boundary as raw JSON rather than the base64
a proto bytes field renders as, because the dashboard sends it straight back as
the replace payload.
ResourceStore is exported from the controlplane package and the State Store
composition asserts against it, replacing the inline anonymous interface that
had started to drift from the real requirement.
The control plane replaces this, rather than extending it
(docs/prds/runtime-control-plane.md, "First implementation"). Until now both
paths existed: overlay kept dashboard edits in its own SQLite file beside
archie.db and layered them over the file config at boot and on every reload,
which is the duplicate configuration authority the PRD set out to remove.
Gone with it: the --no-config-overlay flag and ARCHIE_SKIP_CONFIG_OVERLAY
recovery hatch, which existed because a bad overlay row could brick the boot;
ReloadStatus.OverlayUnavailable; and the reload controller's overlay source.
The -config-overlay file overlay is a different thing and stays.
The dashboard's three write seams went too. UpdateConfig, ResetConfig and
UpdateRepoField were assigned only in internal/webui's own tests, so in
production they answered 503 and ConfigView.Editable was permanently false
(archie-core-j28m). Removing Editable would otherwise have flipped the SPA's
controls on, since configEditable read a missing field as true, so the config
pages become what they already were in practice: a read of what each process
is running with. Configuration is written through the control plane.
apply.go survives this deliberately; archie-core-xrux carries the question of
whether its map merge is still guarding the file overlay path.
The Web UI and messaging channels now read and write Archie's runtime
settings through one versioned, audited control-plane API hosted on the
State Store's existing gRPC server: catalog, query, command, watch. Each
feature owns its resource, validation and commands; storage only writes
versions and audit records.
Workflows move with it. Definitions become portable YAML compiled through
a closed, typed step registry and stored as a control-plane resource, so
the yaegi interpretation paths (internal/gate, wfeval, wfextract,
gateeval, gateextract, skillbuild) are deleted rather than carried.
The UI process reads definitions as data: the JSON projection lives in
internal/domain/workflow/task so a dashboard decodes the stored
collection without linking the workflow engine, which cmd/archie-ui's
deletion gate forbids. Shipped definitions ride the catalog descriptor
rather than a bespoke REST route, keeping the API to its four operations.
archie-messaging gains -state-store-target/-state-store-token and reads
channel settings from the database, failing closed when the State Store
is unreachable: the database is authoritative for those settings.
The pane was capped against the viewport, which let it grow taller than the
box it lives in on a short list -- and a sticky element taller than its
container is a sticky element that cannot stick. It is now capped by the
list's own measured height and by the room the window has, whichever is
smaller: both are sizes the pane does not influence, so filling the cap
cannot feed back into it. Stacked, where there is no sibling to be bound by,
the panel caps itself at 62vh the way the task log does, so one large
payload is a scrolling payload rather than a ten-thousand-pixel page.
The capture detail was the app's only sheet: it covered the list it was
opened from, had to be dismissed between rows, and held no address. It is
now a pane that sits with the list -- sticky and internally scrolling on the
wide layout, stacked under the list below 1100px, where selecting scrolls it
into view. The selection rides in the URL (?capture=N) with replace, so a
capture survives a reload and can be linked, and a page opened without a
selection shows the newest event rather than an empty box.
Posts to /webhooks/capture/{source} with a realistic GitHub, Sentry, form or
large payload (or an exact body), reads the capture back so a silent drop is
visible immediately, and names the intake's rate and size limits when it
hits them. Wired as task webhook:send.
A keyed read whose request never settles pinned the panel's loading line
forever; after a quiet minute the read is declared broken so the retry
button appears, and the started-dedupe is released so it can run again.
Also drops the download button from the log's empty state -- a log with
nothing recorded has nothing to download -- and the leftover imports from
the empty-state cuts.
A title that states the fact does not need a description that derives it:
the log, panels, lifecycle lists, dashboard cards, stages, skills and
captures empty states now say the fact once, and keep only the
descriptions that carry an action, a cause, or a distinct reading.
The selects' first options already say All levels / All stages, so the
uppercase captions restated them; the screen-reader names stay. The
no-stage-names notice restated the filter's obvious emptiness and is gone.
The head strip breaks into a full-width control row -- Back, Refresh, and
Start a new run on the left, the forge coordinates pushed to the far edge
-- and the meta row pairs the lifecycle badge with the attempt pager, which
rides in from TaskMetaBar's slot. TaskMetaBar narrows to the served status
badge; the forge links move out of it into the control row where the other
navigation lives.
The attempt selector becomes an icon-ended pager from the registry
pagination primitive: chevrons walk run history, numbers land on an
attempt, and the active page is the attempt on screen. It sits top-right
above the tab card, and the meta line loses the attempt text fact since
the control now carries it.
The page stacked five chrome strips before its content and every state
wrote an essay. The head is now one strip -- title, workflow, actions, a
single meta line -- the retry explainer is gone (the confirm dialog already
carries the commit-discard consequence), and empty states say the fact in
the title without a paragraph re-deriving it.
The bright violet passed the contrast ratio on paper and failed in the
hand: black-on-purple vibrated at button and FAB sizes. The primary surface
is now deep enough for white labels, and link text on the canvas uses a
dedicated lighter tint, since one violet cannot be both a dark-enough
surface and a light-enough text colour. Also drops an empty-state
description that restated its title.
Ctrl/Cmd+K opens a palette over the capability-gated nav entries; the tasks
board refreshes from the shared event stream; config tooltips are
keyboard-reachable; live activity announces politely; and missing event
timestamps render an em-dash instead of a fabricated now.
Task rows render one primary control with the rest in an overflow menu
instead of every catalog action at once; dangerous-action approvals and the
daemon-restarting install confirm by name; the Needs-you tile reads the
served attention statuses instead of a hardcoded pair; and dangerous.go
stops calling the 24-hour approval permanent.
Empty states and card descriptions no longer explain the consequence the
main clause already implies, page descriptions match between the router and
the chat page-index tool, and the Appearance cards lose descriptions that
restated their controls.
Two things reported from use.
The topbar dropdowns shifted the whole page on click. reka's DropdownMenu is
modal by default, which locks body scroll; losing the scrollbar widens the
layout by its width (measured: 15px at 1440), so everything moved sideways. It
also blocked clicking the page while a menu was open. Non-modal is now the
default for every dropdown, not just the nav's: a menu is not a dialog.
Hover descriptions became a preference on /system/appearance, applied in the
Tooltip primitive so a tooltip added later honours it without each call site
remembering. On by default; nothing depends on a tooltip to be usable.
Appearance also lost its duplicated copy: the page, its first card, and that
card's field each said "Appearance" or "Theme" in turn.
The transcript scrolled itself with a watcher that set
`scrollTop = scrollHeight` on every new message and streamed token, reaching
into ScrollArea's internals through `[data-slot="scroll-area-viewport"]` to find
the element that actually scrolls.
That follows the tail and nothing else. It has no notion of the reader having
scrolled away, so every streamed token yanked them back, and no way down for
someone who had deliberately scrolled up.
shadcn-vue's MessageScroller is the component for exactly this, so the surface
is composed from it instead: the provider owns autoscroll, the viewport reports
its edges, the content is a `role="log"` region, and the item carries the scroll
anchor. It buys the part the watcher could not express: once the reader
scrolls, autoscroll stops rather than fighting them, and a "Scroll to end"
button appears.
Verified in the browser: at the tail the root reports `data-scrollable="start"`
and the button is inert; after a wheel scroll it reports `data-scrollable="end"`,
autoscroll disengages, and the button is visible at opacity 1 and reachable. A
programmatic scrollTop does not count as intent, which is right: autoscroll
reclaims the tail unless a person asked for something else.
Every page from the deleted Preact dashboard, rebuilt in Vue 3 + shadcn-vue.
The navigation restructure landed earlier on this branch; this is the surfaces
behind it, and the two are now one app.
Dashboard, Tasks (list and detail), Configuration split six ways, Logs,
Inspector, Mappings, Bindings, Workflows, Skills, Curators, Channels, chat.
Components, not files: docs/prds/dashboard-port-brief.md requires one
component, one job, one file, with shared state in a module beside it. So a
page file composes and holds nothing, and a shape repeated across panes becomes
its own component (LogRow, PageHeader, PanelLoading, WorkflowBar).
Behaviour that had to survive, because each was a fix:
- Live activity's detail is clamped to one line with the payload on the title.
Unclamped, one row made the dashboard 7,481px tall.
- Each health card takes its own height. Equalising left a mostly-empty box
taking a quarter of the first screen once setup was done.
- Tasks: `tab`/`attempt` live in the query string and are written with
router.replace, so a tab click adds no history entry and does not remount.
The list's deep-link reveal honours prefers-reduced-motion.
- LOG_LEVELS is a wire contract (the server matches that CSV server-side), used
unchanged by both log panes.
- The chat panel's Phase 5 list: a FAB that is the only toggle, a morph whose
scale, radius and position interpolate together, no scrim, no close button,
Escape returning focus to the launcher, Stop only while streaming.
- Unconfigured is not broken. A 501 on /api/version or the log stream renders
what is true ("Not configured on this deployment") rather than blank, and a
read-only Configuration says so instead of offering controls that cannot work.
Three things the port decided rather than carried:
- Button gains an `attention` variant and Badge the five status tones, in the
project's own cva. The pill's dot did not come across: the Badge's shape is
the affordance, and the same tone now covers a trend indicator.
- The icon module is not ported. Its name-to-glyph lookup is the pattern
shadcn-vue's rules call out as wrong, so icons come from @lucide/vue at the
call site. The five bespoke glyphs have no consumer yet and become components
if a page draws them.
- The theme control left the topbar for /system/appearance, per the approved
navigation design. ThemeToggle.vue and Placeholder.vue are deleted rather
than orphaned: every route now resolves to a real page.
Also fixes the bundle freshness check failing on the first build after every
change. ui/dist is committed, so nothing ignores it and Tailwind's content
detection scanned it, making each build an input to the next.
The control room, on the primitives from the previous commit. Being the first
page it also settles the shape of the rest: shadcn components for chrome, one
feature folder per page, Lucide imported at the call site.
Carried across deliberately, because each was a fix rather than a layout:
- Live activity's detail is cut to one line and clamped again in the cell, so a
long unbroken token (a path, a URL, a hash) cannot widen the column past the
table. Unclamped, one agent_finish row made the dashboard 7,481px tall.
- Each HEALTH card takes its own content height. Matching them was right while
both held a checklist; once setup is done that card is a title and a line
while Gate pulse carries a gauge and four rows, so equalising them left a
mostly-empty box taking a quarter of the first screen.
The task links are router paths now, not the `#/tasks` hashes the Preact build
used. History routing means a hash is not a route any more, so one that
survived here would navigate nowhere.
Also in this change:
- Button gains an `attention` variant, in the project's own cva. An action that
exists because something needs a human has to look unlike one that is always
present, and `destructive` would overstate it: nothing is broken, it is
waiting.
- The setup panel branches on the state machine rather than on "is setup
present at all". Guarding on presence showed a 100% checklist with every step
struck through where the complete state belongs.
The Preact tree's ui/src/base/ was deleted with the dashboard. This rebuilds
what depended on it, on the shadcn layer rather than as new bespoke markup.
Badge and Card come from the shadcn-vue CLI, so they match the already-vendored
components, and the composed pieces sit on top of them:
- Badge gains the five status variants the pills used, over the palette's status
tokens. The pill's dot did not come across: the Badge's own shape is the
affordance, and the same variant now covers a trend indicator, where a dot
reads as noise.
- base/StatTile.vue uses full Card composition instead of dumping everything
into one content slot, with base/Sparkline.vue and the goodDirection logic
that says whether "up" is good news.
- base/Gauge.vue and base/SegmentBar.vue, both of them: the original exported
two components from one file and budget composition uses the second.
- base/LogRow.vue over lib/log.ts, which holds the LOG_LEVELS wire contract
(the server matches that CSV server-side, so it is not a display choice).
- lib/status.ts, so status colour has one home: a task state, a gate result, a
log level and a trend all draw from the same five kinds.
Two deliberate departures:
The icon module is not ported. Its name-to-glyph string lookup is the pattern
shadcn-vue's rules call out as wrong, and nothing consumes the five bespoke
glyphs anyway: the navigation is text with tooltips. Icons come from @lucide/vue
at the call site, and the five become components if a page draws them.
The gauge's aria-label announces the rounded percentage. It announced the raw
one, so a screen reader read 71.38461538461539% where the screen showed 71%
(bead 86au).
Also lands the four modules the checklist marks copy-verbatim, which encode
bugs fixed just before the delete: dashboard/activity-detail, chat/
command-scroll, workflows/workflow-rows and logs/logs-empty.
Two Vue branches were in flight and looked like competing versions of the same
change. They were not: this branch had the navigation restructure (Phase 4)
and the route table, and the other worktree had the Phase 1 foundation, still
uncommitted. Neither had ported a page. This reconciles them.
What came across from the foundation:
- lib/api.ts, the one client every request goes through, so the CSRF and
Content-Type headers live in a single place, plus the four framework-free
modules it needs: uuid, stream-state, format, task-meta.
- loadTaskMeta() at boot, so /api/task-meta replaces the freeze-dried
lifecycle defaults instead of the catalogue being frozen into the bundle.
- capabilities.ts keeps the other branch's pure hiddenRoutes, but reads
route.meta.section. The other version read route.section, which is always
undefined for this route table, so its capability gating was silently dead.
The route table stays this branch's: internal/gateway's registry test holds it
in parity with the agent-facing page index in both directions, and the other
branch's table fails it with seven errors.
The frame is Tailwind on the shell, not a ported stylesheet. layout.css
described a shell, a topbar and a nav as authored classes, but the topbar is a
Vue component like every other surface and emits none of them, so two thirds
of that file had no consumer. What the frame needs is now the shell element's
own utilities plus the canvas, focus floor and motion preference in
style.css's base layer, where a global element rule belongs. Verified equal to
the stylesheet it replaces: 20px radius, hairline border, blur(28px)
saturate(1.2), and the narrow-screen collapse.
Also fixes the bundle freshness check failing on the first build after any
change, which had cost both branches of this port a rebuild-and-retry. ui/dist
is committed, so nothing ignores it and Tailwind's content detection scanned
it, making every build an input to the next: class strings left in the bundle
generated utilities for classes nothing references any more, so an unchanged
tree built a different bundle. Excluding it also drops ~12KB of dead CSS.
Takes style.css, lib/theme.ts and index.html's pre-paint script from the
palette work on worktree-groovy-imagining-widget (73e602a7..72034249), leaving
its flat topbar and route table behind. The theme is a data-theme attribute set
before the first paint, so a reload does not flash the wrong palette, and
color-scheme moves with it.
The outlet takes the viewport height below the topbar so an empty state has
somewhere to sit.
Five top-level items instead of eleven flat ones: Dashboard as a link, then
Work, Agent, Events and System as dropdowns, with each item's one-line
description on a hover tooltip. Configuration dissolves into /system/{status,
appearance,tasks,models,repos,advanced} and /settings redirects to the first of
them, so existing bookmarks still land somewhere. Logs gets a nav entry for the
first time.
The bar is composed from named components rather than one file: Brand (Logo,
BrandName), Nav (NavTrigger, NavGroup, NavItem), SearchBar, SupportLink,
ThemeToggle, ProfileDropdown. The nav model reads the route table, so a route
and its nav entry cannot disagree, and a group whose every child is gated off
is dropped while a group left with one child stays a dropdown.
The Go page registry moves with it, which its parse-the-route-table test
enforces.