wyuc 04621578de
release: OpenMAIC 1.0.0 — the agent workbench (#1228)
* feat(storage): add an agent-session store with PG backend and layered contracts (#1163)

* feat(storage): add agent-session store with PG backend and layered contracts

* test(storage): avoid BigInt literals for pre-ES2020 root typecheck

* fix(storage): close agent-session store review findings

* docs(storage): align hook ordering and contention-probe claims with the code

* ci: run on the agent-workbench integration branch

* chore(storage): bump to 0.5.0 for the agent-session store

* fix(storage): carry replay compaction across page boundaries

* feat(agent): add the driver model contract and stage route dialect (#1165)

* feat(agent): add the driver model contract and stage route dialect

* fix(agent): validate route context windows and clarify dialect precedence

* feat(agent): adapt the agent-session store and runtime foundations (#1167)

* feat(agent): adapt the agent-session store and runtime foundations

* feat(agent): resolve request owner identity via an anonymous cookie

* docs(agent): document the opt-in compaction default and harden edge cases

* feat(agent): add the background session runner (#1169)

* feat(agent): add the background session runner

* feat(agent): wire the runner into startup behind feature flags

* fix(agent): stop clean interruptions from consuming the attempt budget

* fix(storage): charge the attempt budget for abandoned leases but not clean parks

* docs(storage): document the attempt-charging contract and decouple its tests

* feat(agent): add agent session and owner event streams (#1170)

* feat(agent): add agent session and owner event streams

* fix(agent): close the session-existence oracle and document the owner seam

* feat(agent): add agent session lifecycle routes (#1171)

* feat(agent): add agent session lifecycle routes

* fix(agent): validate session-create input and preserve the owner cookie on errors

* refactor(storage): drop the unused active-stage API from the agent-session contract (#1174)

* refactor(storage): drop the unused active-stage API from the agent-session contract

Tools address stages explicitly on every call, so the store keeps no
mutable session-level stage pointer. Removes resolveActiveStage and
setActiveStage from the store interface, their PG implementations, the
active_stage_changed lifecycle event, the session_active_stage owner
event variant, and the contract tests pinning them. The active_stage_id
column and the DDL check constraint stay untouched for schema
compatibility.

* chore(storage): bump @openmaic/storage to 0.7.0 for the contract removal

* docs: document the agent runtime configuration surface (#1176)

* fix(agent): repair orphaned and late tool results across interruption boundaries (#1180)

* fix(agent): repair orphaned and late tool results across interruption boundaries

A crash, shutdown, or provider failure can leave the durable transcript with
tool calls that have no result, or with results ordered illegally for the
provider. Three failure modes were fixed:

- Orphaned tool calls: a run that died between an assistant tool-call frame
  and its result left a dangling call in the entry tree. Resume no longer
  synthesizes and persists receipts for it: interrupted results are a
  read-time provider view owned by a shared read-boundary repair, which
  returns the original array for a healthy transcript and never mutates the
  tree.

- Late parallel results: a parallel tool can finish while pi unwinds an
  aborted assistant frame, leaving result(A), assistant(aborted), result(B)
  in durable order. Strict providers reject non-contiguous results, so the
  read-boundary repair moves existing results next to their owning assistant
  frame (in call order), omits incomplete unwind frames, and synthesizes
  receipts only for genuinely missing calls.

- Interrupted calls at the write boundary: a call still in flight when the
  run winds down (shutdown, lease loss, cancellation, provider failure) had
  no receipt at all. The runner now tracks in-flight calls from their
  assistant frames and, before the terminal flush, appends an
  interrupted-result receipt for each still-orphaned call through the same
  attempt-fenced write chain, so a lease-stealing zombie never writes and
  the next claim sees a provider-safe transcript.

* test(agent): pin the runner wiring for interruption-boundary tool repair

* feat(agent): add neutral tool foundation libraries (#1184)

* feat(agent): register a web_search tool on the session runner (#1185)

* feat(storage): add a per-session URL trust gate (#1186)

* feat(agent): add the skills system (#1189)

* feat(agent): add the skills system (builtin directories and durable user skills)

* fix(storage): serialize the user-skill quota check-and-insert per owner

Two concurrent creates at the 50-skill boundary both counted 49 rows and
both inserted (READ COMMITTED, no lock), overshooting the quota contract.
The create transaction now takes a per-owner pg_advisory_xact_lock first,
and the same-name idempotency check runs before the count check so an
at-least-once retry of the create that committed as the owner's 50th row
still returns its durable receipt instead of a quota error. The 23505
backstop is retained for writes that do not take the lock.

* fix(agent): share unstorable-character validation and align skill lookup

* feat(agent): add session materials and a fetch_url tool behind the URL trust gate (#1190)

* feat(agent): add session materials and a fetch_url tool behind the URL trust gate

* fix(agent): harden session material fetching

* feat(storage): add an ownership scope to stage documents (#1191)

* feat(agent): add material read and search tools (#1192)

* feat(agent): add stage read and patch tools (#1194)

* feat(agent): add page generation and deck editing tools (#1198)

* test(storage): keep the PG contract suite order-independent (#1200)

* fix(agent): revoke deleted-session URL authority and reject private ISATAP endpoints (#1199)

* fix(storage): revoke deleted session URL authority

* fix(ssrf): reject private ISATAP endpoints in strict fetches

* chore(storage): bump to 0.11.1 for the session-URL authority fix

* feat(agent): add roster and voice registration tools (#1201)

* feat(agent): add folder organisation tools (#1202)

* feat(api): add stage and material HTTP routes (#1203)

* feat(workbench): add the client data layer (#1204)

* feat(workbench): add the client data layer

* docs(workbench): write the ported comments in English

* chore(edit): remove the in-editor agent panel (#1210)

* chore(edit): remove the in-editor agent panel

* style: apply prettier formatting

* fix(agent): report the runtime as unusable without a database (#1207)

* fix(agent): report the runtime as unusable without a database

* style: apply prettier formatting

* feat(agent): add image, video and pptx import tools (#1211)

* feat(workbench): add the agent chat surface (#1205)

* feat(workbench): add the agent chat surface

* docs(workbench): write the ported comments in English

* fix(workbench): label the folder and rename tools on the timeline

* fix(workbench): label the roster and voice tools on the timeline

The reconciliation test iterates every tool the runner registers and
requires a display label of its own. The roster and voice-clone tools
(list_voices, set_roster, clip_audio, register_voice) reached the
integration base with the roster/voice-registration tools but never
gained presentation rows, so they fell through to the default branch
and rendered their wire names. Port their rows from the reference
implementation (labels and i18n keys verbatim) and extend the
reconciliation allowlist with ROSTER_TOOL_NAMES and
VOICE_CLONE_TOOL_NAMES, so a future tool cannot enter the product
without a label.

* feat(agent): add the material extraction lifecycle (#1212)

* feat(storage): add material extraction lifecycle

* feat(agent): execute queued material extraction

* style: apply prettier formatting

* style: satisfy prefer-const in the extraction runner

* test: give material fixtures the extraction lifecycle fields

The media-tools slice and the extraction lifecycle slice were each green
in isolation but never compiled together: the lifecycle made derivedFrom
and extraction required on AgentSessionMaterial while the media-tool
fixtures predate them.

* chore: remove stray task notes

* fix(workbench): label the extraction lifecycle tools on the timeline

* feat(workbench): add the workspace shell (#1206)

* feat(workbench): add the workspace shell

* docs(workbench): write the ported comments in English

* i18n(workbench): align workspace keys across locales

* fix(workbench): adopt the landed data layer and label the extraction tools

- replace the sibling-slice seam stubs with the real data-layer modules
- drop ambient declarations now shadowed by landed files
- port timeline labels for the extraction lifecycle tools from the reference
- align the new i18n keys across all locales

* ci: retrigger

* feat(api): folder routes, stage-meta viewer surfaces, and the material upload contract (#1215)

* fix(storage): restore capability-based stage access

* fix(api): bind document access to request owner

* fix(agent): restore three-state stage access on the tool layer

Port probeStageAccess and the three-state StageAccess (owned / foreign /
missing / tombstoned) and gate every stageId-bearing stage tool on an owned
probe, mirroring the reference per tool:

- move_to_folder, rename_stage, read_stage_outline refuse a non-owned stage
  with the single not-yours message before touching the store.
- The course/DSL toolset and the roster toolset are wrapped by
  withOwnerStageAuthorization: read_stage, patch_stage, grep_stage and every
  writer refuse a foreign stage with the same message and refusal shape.
- Scene preview keeps its own probe and its own refusal text, and is
  registered beside the course toolset (never double-gated).
- The runner injects one probe factory at the three call sites.

Tests: the dsl cross-owner test premise (a foreign stage is readable by id)
encoded an invented capability-read policy that the reference does not have
at the tool layer; it now asserts foreign read/patch/grep are all refused
while the owner still reads. Curriculum cross-owner assertions were already
the reference's and now pass with the probes in place.

* docs: correct per-file test counts in the fidelity report

* test: fix type errors in stage-access fidelity test

* test: adapt media-tool and gate suites to the owner-scoped store seam

* feat(api): add owner-scoped course-folder HTTP routes

Port the reference implementation's /api/folders family (list, create,
rename, delete with ungroup/remove modes, and folder membership) onto the
owner-bound document store, replacing its provider-based auth with the
existing withRequestOwnerId / owner-scoped store seams.

The storage package's folder store grows the pieces the routes need:
DocumentFolder.order (schema column + max+1 assignment + ordering),
renameFolder, deleteFolder(mode) with captured member ids, and
setStageFolder(stageId, folderId | null) with idempotent un-filing.
FolderNameError moves into folder-name-validation.ts (stage-storage re-exports
it, keeping import sites intact).

Every route gates on the configured agent runtime (plain 404 when off or
unconfigured), keeps the reference's machine codes and envelopes, and is
covered by gate tests plus a behavior suite.

* feat(api): add stage-meta viewer surfaces for the classroom

Port the reference implementation's viewer-facing stage state — can-edit /
collected / published / generation-complete — on top of the stage-access
base (stage_meta + tombstones). stage_meta gains published_at and
generation_complete columns plus a stage_bookmarks table; the reference's
deployment-specific origin/claimed_at columns are stripped.

New gated routes: GET /api/stage-meta/[stageId] (per-viewer facts, 404 for
absent/tombstoned, never returns the owner id), GET /api/stages/[id]/status,
POST generation-complete / publish / unpublish (owner-only), POST
/api/bookmarks. The resolver lives in lib/server/stage-access.ts.

Wiring: a fetchStageMeta client with the reference's three-outcome contract,
stage-store isOwner/isBookmarked/readOnly fields (upstream single-user
defaults, no-op until the sidecar answers) plus setViewerAccess, the
classroom apply path computing readOnly = !(isOwner || isBookmarked), the
Stage editability gate, and a sidecar probe after each classroom load. A
sidecar 'absent' answer keeps the editable default here because the
classroom also serves local-only courses; server writes stay owner-enforced.

* feat(api): port the reference material upload contract

Rewrite POST /api/materials to the reference implementation's upload shape
so the workbench uploader (uploadWorkbenchMaterial, which posts no session
id and expects a flat 201 view) works unchanged: owner-scoped upload with
mime normalization/validation (415), per-class size caps checked on the
declared content-length and the streamed body (413), empty body (400),
quota (429), sha256 reserve->store->finalize lifecycle with abandon on
failure, flat { materialId, originalName, bytes, mime, extraction } 201,
and an x-request-id echo.

Adds the owner-scoped material library (owner_material table + quota +
24h lazy sweep, bytes in the host's asset registry as the neutral
replacement for the reference's object-storage byte path) and the material
cap configuration. The session-scoped GET list is left as-is; the
reference's owner-material extraction worker is not ported (the branch's
session-material extraction lifecycle already covers extraction).

Gate tests now cover all 23 persistence routes across the three runtime
env states; the materials behavior suite pins the new contract.

* feat(media): add an optional local ffmpeg media extractor (#1213)

Adds a local ffmpeg/ffprobe pipeline as a second media extraction
provider behind the extractor registry, ported faithfully from the
reference implementation: duration probing, keyframe-safe chunking,
per-chunk ASR with timeout and deadline budgets, and timestamped
transcript assembly.

- Availability probing feeds the registry's candidate selection: the
  provider simply is not a candidate when ffmpeg/ffprobe are absent.
- With neither ffmpeg nor a cloud provider configured, extraction fails
  with an actionable message naming both enablement paths.
- Media materials route through the same extraction lifecycle and lease
  fence as documents; no parallel queue.
- Tests inject the executable resolver so the missing-ffmpeg path is the
  default-tested one; the real pipeline test is skip-if-unavailable.
- @openmaic/storage 0.13.0 -> 0.14.0 (media routing in the material
  lifecycle surface).

* feat(storage): per-scene monotonic revisions via database triggers (#1214)

* feat(storage): per-scene monotonic revisions via database triggers

Restore the reference implementation's freshness granularity: a
per-scene monotonic revision maintained by database triggers, so every
writer (HTTP routes, agent tools, jobs, manual SQL) bumps it without
application cooperation.

- Companion revision tables + trigger functions in the storage package's
  idempotent schema bootstrap, with the lock-order invariant, pg_notify
  wakeup and the suppression switch for batch writers.
- ensureDocumentSchema gained a dollar-quote-aware statement splitter.
- The freshness and manifest routes serve per-scene revisions.
- Mutation-verified: dropping the triggers turns the revision tests red.
- @openmaic/storage 0.13.0 -> 0.14.0.

* fix: forward the freshness manifest through the owner-bound store

* feat(workbench): add the Pro entry points and preserve the mode-transition semantics (#1208)

* feat(workbench): add the Pro entry points

* feat(workbench): preserve Pro mode transition semantics

* fix(workbench): drop ambient declarations shadowed by landed slices

* fix(workbench): drop ambient declarations shadowed by the landed shell

* feat: port workspace shell sibling modules

Port the 16 leaf modules the Pro workspace shell imports but that were only
ambient-declared, replacing the compile-time bridge with real implementations
adapted from the sibling-slice reference: pure workbench helpers (session
title, rail tab, course-chat bootstrap, created-course tabs, course-tabs
memory, workspace navigation, pane navigation, pro-edit sizing,
existing-course minting, first-message session), the neutral brand context and
course-rename server API, the server-action session delete, the home
discovery hook, the classroom pane host with its load-policy leaf, the theme
toggle and floating-layer owner, plus the floating-layer-owner wiring the
dialog/dropdown/tooltip portals stamp.

Also add the workbench-shell locale copy for all 12 locales, port the
reference tests for the ported modules, and drop
types/workbench-sibling-slices.d.ts now that every declaration has a real
implementation.

* docs: keep ported comments in English and deployment-neutral

* docs: announce 1.0.0 and refresh the feature overview (#1216)

* docs: announce 1.0.0 and refresh the feature overview

* docs: finalize 1.0.0 README after feature merge

* fix(agent): control-plane routes answer 404, not 500, without a database

The agent control-plane routes gated only on the runtime flag, so an
enabled-but-unconfigured deployment (flag on, DATABASE_URL empty)
answered 500 from a store that cannot connect. Gate them on the
configured check instead, matching the stage/material routes: the
whole surface is cleanly absent until both the flag and the database
are present. The status probe keeps reporting both bits.

* test: mock both runtime gate exports in the control-plane route suites

* fix(agent): abort in-flight TTS on cancel and bound each provider request with a timeout (#1217)

The generate_tts / scene-tts path checked the runner's AbortSignal between
actions but never created the provider HTTP requests with it, so a session
cancel left a hung synthesis fetch in flight until a restart repaired the
tool result. Thread the signal end-to-end: TTSModelConfig carries an
optional signal, generateTTS combines it with a per-request timeout
(TTS_REQUEST_TIMEOUT_MS, default 30s, ported from the reference runtime's
TTS bounds) via AbortSignal.any, and every provider fetch (openai, azure,
glm, qwen incl. voice-clone + audio download, voxcpm, minimax, doubao,
elevenlabs, lemonade) is created with that signal.

A timeout now fails the tool call with TTSRequestTimeoutError (a clear
retryable error) instead of wedging the session; a caller cancel propagates
as the interruption so the runner settles the session as cancelled without
a restart.

Tests: hung-provider simulation rejects at the timeout with the retryable
error; abort mid-flight aborts the captured request signal and surfaces the
interrupted shape; removing the signal wiring makes the abort tests fail
(red), restoring them turns green.

* fix(workbench): PG-mode home listing via owner stages; keep the interrupted terminal course card (#1218)

Finding 1: with server persistence on, listStages resolved to the generic
GET /api/persistence/documents listing, which the capability model
deliberately answers 403 FORBIDDEN_DOCUMENTS for (reads by id, listings
owner-only). The home/workspace library now lists through the owner-scoped
GET /api/stages surface (same anonymous-owner cookie the workbench uses)
when server persistence is enabled; the server-side 403 is untouched.

Finding 2: a run interrupted (session_interrupted) and repaired
(session_resumed) that ends cancelled before agent_end stranded its pending
classroom sightings, so the timeline's terminal card lost the course the
answer produced. session_end (cancelled) now flushes the pending sightings
into the same course card set agent_end paints, before the stopped caption.

* chore(workbench): remove the bookmark concept and the saved-courses drawer (#1219)

* chore(classroom): remove the bookmark ('collected') concept entirely

The stage-meta viewer port introduced a bookmark surface (stage_bookmarks
table, POST /api/bookmarks, the isBookmarked sidecar field, and a
readOnly rule that let a saved course stay editable). The product has no
such concept, so remove it as a closure:

- delete the /api/bookmarks route and the stage_bookmarks table plus its
  query helpers from the persistence bootstrap
- drop isBookmarked from GET /api/stage-meta/[stageId]
- simplify the classroom read-only rule to readOnly = !isOwner across the
  sidecar client, ownership signal, classroom load, stage store and the
  classroom page
- keep publish/unpublish, generation-complete, isOwner and isPublic
  exactly as they were
- update the gate and stage-meta route suites and the README mentions

The workspace rail's Bookmark glyphs and comments describe the upstream
saved-courses (favorites) section, which is driven by isOwner and renders
no collect affordance; they are kept as unrelated homonyms.

* chore(workbench): remove the saved-courses drawer UI

The first pass removed the bookmark data model but kept the rail's
"Saved courses" drawer, judging it a separate surface driven by
`isOwner === false`. The home/workspace listing is owner-scoped, so that
flag can never occur: `allSaved` is permanently empty and the drawer
(plus the collapsed-rail Bookmark mini-button) is a dead affordance.
Remove it: the SavedDrawer component and its mount, the savedOpen /
savedSection state, the allSaved / matchedSaved derivations, the 'saved'
variant of the course-list renderers, the mini Bookmark glyph, the
drawer-only CSS, and the drawer's i18n keys from all 12 locales. The
courses tab is now exactly one folders tree.

The authored/favorites split in workspace-tree.ts goes with it; the tree
module no longer reads `isOwner`. The discovery course type keeps the
field — the shell still reads it for read-only gating.

Upstream has no collect concept; the drawer could only ever render empty
here. The reference implementation HAS this drawer (its favorites come
from its account system), so this removal is a deliberate upstream
product decision, not a fidelity bug.

* fix(workbench): restore the attach entry, add the rail settings entry, pin all three entry points (#1221)

* fix(workbench): restore the composer attach entry by gating it on the live runtime

The AttachButton's rollout probe read a `materialsEnabled` field that this
branch's /api/agent/runtime never answers (the materials routes gate on the
runtime itself, like the stages), so the gate could never pass and the attach
button never rendered — the Pro launch and chat composers showed only the
@-mention and enhance glyphs.

Substitute the field with the runtime's `enabled` value, which IS the upload
action's precondition: POST /api/materials answers 404 whenever it is false,
so the render condition now equals the action precondition (no dead button).

The button's label (`proMode.attach`) is a user-visible string that becomes
visible again; port the reference implementation's own translations verbatim
into the 11 locales that still carried the Chinese copy.

* feat(workbench): add the settings entry to the rail's bottom-left cluster

The reference's rail foot carries a cluster of utilities (its saved-courses
drawer, the language switcher, the display toggle). This branch removed the
drawer — it could only ever render empty here — and the product decision is
to fill that freed spot with the settings entry.

Add a settings trigger to the foot cluster (expanded rail, beside the
language and display toggles, and on the collapsed strip) and mount the
model/provider SettingsDialog in the rail, wired to the trigger. It is the
same dialog the classic home opens from its header pill; the workspace had no
settings entry of its own, so nothing is duplicated within a surface.

* test(workbench): pin the restored upload, attach, and settings entry points

Covers the three restored entry points:

- the courses-tab upload control: rendered beside the course name filter,
  wired to the discovery hook's ZIP import trigger, disabled while an import
  runs, and gated by the same condition as its action (the courses tab);
- the composer attach control: an actual render of AttachButton under both
  probe answers (visible when the runtime says the upload path is live,
  hidden otherwise), its mounts in the launch and chat composers, the
  branch's runtime-field substitution in the probe, and the reference's own
  `proMode.attach` copy in all 12 locales;
- the settings entry: the trigger in the rail's foot cluster (expanded and
  collapsed), beside the language and display toggles, opening the
  SettingsDialog the rail mounts.

* chore(config): the Pro workbench flag implies the MAIC Editor gate (#1223)

A workbench build without the editor toggle has no way to edit a course:
enabling NEXT_PUBLIC_PRO_WORKBENCH_ENABLED while forgetting
NEXT_PUBLIC_MAIC_EDITOR_ENABLED produced exactly that split-brain bundle.
The workbench IS Pro mode, so its flag now implies the editor gate; the
standalone flag remains for deployments that want the classroom editor
without the workbench. Documents both flags in .env.example.

* fix(agent): wake SSE tails and the runner on durable deltas (streaming fidelity) (#1222)

The Pro workbench chat did not stream: the session/owner SSE routes polled
the durable event log on a 5s/30s clock with no wakeup, so message_update
deltas (written at 150ms cadence) reached the browser in poll-sized blocks
and the thinking strip only mounted after the whole reasoning text had
accumulated.

Port the reference's LISTEN/NOTIFY delta path:

- storage: add in-transaction wake hooks (onSessionEventAppended,
  onOwnerEventAppended, onCancelRequested) so a host queues pg_notify in
  the same transaction as the durable append; align readEventsAfterForReplay
  to rank the bounded page so the first delta after the cursor is always
  kept (the live tail can never starve). Bump @openmaic/storage to 0.18.0.
- app: port the process-wide event-notify bus (dedicated LISTEN client,
  self-check probe, reconnect backoff; notify through the storage
  transaction surface), wire the store hooks, subscribe both SSE routes
  before the initial read with the reference's initializing gate, and give
  the runner one {kind:'session'} subscription whose wake runs the cancel
  check and the message drain. Polls stay as the lossy-NOTIFY backstop.
- lifecycle: start/stop the bus from instrumentation.

Tests: storage hook + compaction contract; route wakeup latency; runner
wakeup wiring with a fake agent; bus unit tests; PG contracts proving a
real append wakes the routes and a live SSE route forwards a message_update
on the wakeup, and that a rolled-back append never wakes. Also fix the
pre-existing park-attempt-budget PG test TRUNCATE (missing CASCADE against
newer FK tables).

* fix(storage): asset writes self-deadlocked against pooled PostgreSQL (#1225)

* fix(storage): refuse the non-transactional byte-write deadlock configuration

A byte store whose plain write() runs on its own pooled connection cannot be
invoked from inside a registry write transaction: after the transaction has
claimed the blob-row lock, that write blocks on the lock the transaction just
took while the transaction waits on the write - a self-deadlock PostgreSQL
cannot detect (one side is idle in transaction).

There is no lock-safe ordering for such a writer: bytes must be written after
the row claim (writing before it lets the collector delete the bytes while the
upsert waits), and any second-connection write after the claim is the
deadlock. The configuration is therefore detected and refused:

- AssetByteStore gains writesOutsideRegistryDatabase?: true, declaring that the
  layer's plain byte operations cannot contend for the registry's row locks.
- PgAssetStore refuses put()/replace() up front (and defends coordinatedWrite)
  when the byte store has no writeWith and does not declare the flag, throwing
  a clear configuration error before any row is claimed.
- The collector mirrors the guard on its delete path (deleteWith or a declared
  out-of-registry layer, else a configuration error).
- The object store declares the flag (its out-of-transaction write remains
  legitimate); the in-registry PostgreSQL byte column provides writeWith /
  deleteWith instead.
- Write transactions (put/replace/remove) set SET LOCAL lock_timeout = 30s so
  any future lock-contention variant fails loudly instead of hanging.

Bumps @openmaic/storage to 0.18.0.

* fix(persistence): forward the transactional byte methods through the lazy asset byte-store wrapper

The no-bucket case of lazyAssetByteStore returned a bare { write, read,
delete } and dropped writeWith/readWith even though the underlying
PgAssetByteStore has them. The registry's hasTransactionalWriter duck check
then failed and put() fell back to the byte store's own pooled connection,
which blocks forever on the blob-row lock the registry transaction just took
when the bytes live in the same PostgreSQL - the production self-deadlock.

The no-bucket layer is statically PgAssetByteStore, so its transaction-pinned
methods are forwarded eagerly (typed against the real signatures via
PgForwardedByteStore). The bucket case keeps its lazy-probing semantics: no
transactional writer exists there, the signed-URL method stays absent or lazy
exactly as documented, and the wrapper now declares writesOutsideRegistryDatabase
so the registry may run the plain write inside its transaction.

New tests pin the wrapper's transactional capability red-to-green and assert
put()/resolve() route byte traffic through the transaction-pinned queryable.

* fix(home): cap the generate-prep ingest drain at 3s so Generate never waits the full server budget

The classic home flow's Generate click drained in-flight ingests for the full
15s server budget. Cap the wait at GENERATE_DRAIN_CAP_MS (3000ms, documented as
a UX bound) and reuse the existing timeout fallback: sources that miss the cap
proceed on the legacy byte path and each late-resolving id is released.

* chore(storage): bump to 0.19.0 over the concurrently landed 0.18.0

* fix(agent): bound every tool call with a timeout; never resurrect a cancelled session (#1226)

* fix(agent): bound every tool call with a global timeout and settle it on cancel

A tool await that neither resolves nor rejects wedges the session forever:
the lease keeps heartbeating and the driver never reaches its next cancel
checkpoint. Race every tool execution (in buildAgent) against a hard budget
(OPENMAIC_AGENT_TOOL_TIMEOUT_MS, default 10 min, per-tool overrides for
known long runners) and against the caller's AbortSignal, so even a
signal-ignoring await cannot keep a cancelled session running.

On timeout the call rejects with AgentToolTimeoutError; the agent loop turns
the rejection into a structured error tool-result the agent can retry or
proceed from, and the abort signal is delivered to the tool's in-flight work
through a derived controller. Zombie-tool updates after settlement are
dropped.

* fix(storage): never re-lease a cancel-requested session; settle it as cancelled on claim

The claim scan treated a session with cancel_requested_at set as a normal
claim candidate: after a restart it re-leased the same session for attempt
N+1 and resumed generating despite the pending cancel. claimNextSession now
settles such candidates as cancelled under the claim lock (status cancelled,
attempt reset, lease and cancel request cleared, terminal session_end event
and owner projection) instead of leasing them, then keeps scanning.

Bump @openmaic/storage to 0.18.0.

* docs: takeaway-style 1.0.0 announcement with bilingual guide links

The 1.0.0 head is now a short takeaway block — badge links to the
official user guides (English and Chinese), five one-line highlights,
and pointers into Features and the workbench setup section — instead of
six dense paragraphs. The detailed provider-neutrality and freshness
notes move into the Features workbench section, phrased database-
neutrally (the announcement no longer names a specific database).
Release date corrected to August 27.

* fix(workbench): restore editor chrome, mode transition, streaming, materials, mentions, folders (#1229)

* fix(workbench): wire workspace folder routes

* fix(editor): restore reference workbench chrome

* fix(workbench): persist composer materials and course refs

* fix(workbench): preserve live reasoning frames

* fix(persistence): back off failed streaming saves

* chore(workbench): retire stale slice seams

* test(editor): cover element pin layer

* chore(storage): bump to 0.21.0 for the user-message ref/material fields

* chore(editor): translate ported code comments to English

* fix(agent): fence durable tool writes and consume cancel requests atomically (#1230)

* fix(agent): enforce provider force-off in agent tools and scrub vendor identity from tool results (#1231)

* fix(materials): serialize per-owner quota reservations and make crashed uploads reclaimable (#1232)

* fix(editor): resolve dock-bar i18n keys, remove dock height drag, wire element referencing (#1233)

* fix(workbench): send the opening session message exactly once with refs intact (#1234)

* feat(editor): port timeline TTS preview single-flight and voice-all state latching (#1235)

* fix(media): restore the reference classic media chain (#1236)

* fix(import): adapt imported PPTX canvas size so decks render without overflow (#1237)

* fix(editor): complete element referencing — renderer DOM contract and GenUI picking aligned with the reference (#1238)

* test(providers): reconcile the provider-config vendor-token debt count after the main merge

The integration line's AK/SK fallback for the managed document provider adds
occurrences that main's allowlist snapshot predates. Same mixed-composition
debt category the group already documents; no new vendor behavior.

* test(providers): reconcile vendor-token debt counts with the integration line

The main-merge brought main's neutrality-guard snapshot next to integration
features it predates (media-extractor fallback chain, local voice-profile
deletion semantics, the enabled-TTS helper). Same debt categories the guard
already documents; counts updated to the guard's own tally and two grouped
entries added. No new vendor behavior.

* fix(agent): carry reasoning through the completions dialect so the thinking strip renders (#1239)

* feat(skills): add Feynman and spiral curriculum methods (#1240)

* feat(agent): port missing reference tools and skills (parity audit) (#1241)

* feat(media): retire asset-registry wiring; media and materials follow the reference byte model (#1242)

* fix(classroom): center adapted canvases in the stage and send back navigation home during generation (#1243)

* feat(settings): skill management with real list, download, delete, and upload (#1244)

* feat(settings): skill management section with real list, detail, and zip download

* feat(skills): owner skill delete and upload across storage, API, and settings

* fixup! feat(settings): skill management section with real list, detail, and zip download

chore: neutralize a reference note in the settings header comment

* fix(media): persist origin-independent classroom-media references from the agent runtime (#1245)

* feat(editor): float the insert toolbar in the outer frame with collapse (#1246)

The insert strip was bounded to the slide card, so it could only ever sit on
top of slide content: the card's overflow clipped it and it could not be parked
in the padding beside the slide. Move it into the studio frame the element
picker's panel already roams (CanvasOverlayPortal + the frame selector), so both
canvas overlays share one bounding container and their handles behave the same.
While picking, the strip rises over the picker and goes inert, which is the
z-order CANVAS_OVERLAY_Z already documents.

Add a fold beside the grip: the chevron collapses the strip to that grip row and
back, with the buttons unmounted rather than hidden. The fold is session-local
state owned by EditShell, next to the drag offset, so a surface swap keeps it;
nothing is persisted. Expanding a strip parked at the bottom edge re-clamps
through the same bounds rule the keyboard move uses.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

* fix(workbench): align the chat timeline's left edge with the composer (#1247)

* fix(agent): fence session claims while an ask_user question is outstanding (#1248)

* fix(agent): settle-time rescue tracks real delivery instead of a count offset (#1249)

* fix(persistence): migrate owner_material to oss_key and drop legacy asset_id (#1250)

* docs(readme): surface the 1.0.0 user guide badges at the top (#1253)

* fix(workbench): show newly created folders in the sidebar without reload (#1254)

* docs(readme): add the release version prefix and drop the opt-in framing

* fix(workbench): single-source the chat gutter so timeline and composer share a left edge (#1255)

The transcript and the composer each established their own column: their own
`px-*` gutter and their own `mx-auto w-full max-w-*` centering wrapper. Equal
padding values were never enough, because the two columns are centered inside
different containing blocks — the transcript's is a scroll container, whose
content box is narrower than the composer footer's by the scrollbar's width:

  transcript text left = pad + (pane - 2*pad - scrollbar - measure) / 2
  composer box  left   = pad + (pane - 2*pad             - measure) / 2

The padding cancels out of the difference and what remains is `-scrollbar/2` at
every padding value, so the transcript sat half a scrollbar to the left of the
composer and tuning the two paddings against each other could not move it.

The column is now established once, by the nearest common ancestor of both
(`chatColumn`), and the scroll viewport and the composer footer are siblings
inside it that add no horizontal inset of their own. The cap carries the gutter
on top of the 760px reading measure, so the text column keeps its width. The
handed-over question row drops the padding that indented it past the agent's
prose; framed rows keep their own inner padding, which is what a card's border
sitting on the column edge means.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

* fix(workbench): lock pane-embedded classroom to edit mode (#1256)

The workspace right pane painted the full learning chrome — speed control,
play button, learner avatars, mic bar — for a course the agent had just
created, then flipped to edit once the first scene landed.

resolveStageChromeMode treated playback as the DEFAULT branch for a hosted
classroom, so every shortfall fell into it: a course whose tab opens at
stage_link time has no scenes yet, so currentSceneId is null and
isHostedSceneEditable is false. A folded pane parked the playback root
behind the fold and cross-faded it out over the pane on unfold, and a
failed editor chunk dropped into playback permanently.

Lock it at the pane instead of defaulting per entry path:

- WorkbenchPanelProvider — the single element that mounts a classroom into
  the workspace — publishes editPinned (visible && !playback). Every entry
  path passes through it, so none of them decides.
- The hosted resolution can no longer degrade to playback. Start Learning
  (workbenchLearning, new input, split out from pane visibility) is the one
  door; everything else resolves between the neutral loading shell and edit.
- Stage's chrome dispatch is exhaustive on chromeMode, so the playback root
  is no longer the else-branch of a condition about the current scene.

No flicker: chromeMode is resolved during render, and preloadEditor now
answers synchronously (isEditorPreloaded) so a remount with the chunk
already registered paints edit on the first frame. A failed import is no
longer cached forever, so the lock cannot strand the pane.

Standalone classrooms keep their stored mode unchanged.

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-27 19:26:25 +08:00

1988 lines
68 KiB
CSS
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

/**
* The Pro workspace's visual layer.
*
* Scoped under `.ws-root` and imported only by `WorkspaceShell.tsx`, so `/`
* never receives a byte of it. Nothing here changes behaviour: every rule is
* colour, type, light, shadow or motion.
*
* INHERITANCE IS THE BRIEF. This is the Pro surface of OpenMAIC, not a second
* brand, so the vocabulary is lifted from what the product already uses:
*
* - the canvas is the homepage's own `slate-50 → slate-100` gradient, not
* the flat `bg-white` v2 shipped (the flat white was the departure);
* - the composer keeps the classic face's chrome — `rounded-2xl`, a
* `border/60` hairline, `bg-white/80` over `backdrop-blur-xl`;
* - the accent is brand violet, spent where the product already spends it:
* focus, active state, the Pro badge;
* - the neutral ramp is slate, the same one `ClassroomCard` and the home
* chrome use.
*
* What is NEW is craft, not vocabulary:
*
* - shadows are layered (contact → mid → wide) instead of one `shadow-xl`
* blur, which is most of why v2 read as unstyled;
* - the composer gets a real focused state — a violet thread, a widened
* violet-tinted cast, and a 1px lift — so the page has one object that
* obviously had attention paid to it;
* - the canvas is lit rather than blank: a very low-contrast violet-white
* pool behind the hero plus a grain wash, both at the edge of visible;
* - rail rows hover into one soft neutral pill and nothing else;
* - one staggered entrance on load, then nothing animates unprompted.
*
* Type stays on the product's Inter stack. The hero's problem was never the
* typeface, it was that 22px/semibold is body copy in a headline's seat; it is
* set properly here (larger, tighter tracking, optical spacing) and left alone.
*/
/* ── Tokens ──────────────────────────────────────────────────────────────
Slate ramp + brand violet, matching the rest of the product. */
.ws-root {
--ws-canvas-top: #f8fafc; /* slate-50 */
--ws-canvas-bottom: #f1f5f9; /* slate-100 */
--ws-rail: rgba(255, 255, 255, 0.72);
--ws-surface: rgba(255, 255, 255, 0.82);
/* The ink ramp, re-cut. Every step used to sit within one slate shade of its
neighbour, so the rail read as one grey band: the counts (slate-300) came
out at 1.5:1 on the rail and the section labels and metadata (slate-400) at
2.5:1 — below AA, which is why nothing announced itself. Each step is now a
full role with a ratio to defend, measured against the rail's own blended
background (#fbfcfd), not against pure white:
ink 16.5:1 course names, folder names, the active row
soft 10.0:1 chat names, section labels — legible, still not black
mute 4.7:1 metadata, counts, glyphs, buttons at rest (AA)
faint 2.5:1 DECORATION ONLY — separators, rules. Never text.
The palette is unchanged: these are all slate steps, just further apart. */
--ws-ink: #0f172a; /* slate-900 */
--ws-ink-soft: #334155; /* slate-700 */
--ws-ink-mute: #64748b; /* slate-500 */
--ws-ink-faint: #94a3b8; /* slate-400 */
--ws-line: rgba(15, 23, 42, 0.11);
--ws-line-soft: rgba(15, 23, 42, 0.065);
--ws-tint: rgba(15, 23, 42, 0.055);
--ws-tint-strong: rgba(15, 23, 42, 0.09);
/* The `@` course menu's row, and how many of them its window shows.
The window is a WHOLE NUMBER of rows so the resting state never cuts one in
half — which is what the old fixed `max-h` did, and with no scroll on the
container that half row was simply unreachable. */
--ws-cmenu-row-h: 30px;
--ws-cmenu-rows: 8;
/* ── The rail's row template ────────────────────────────────────────────
A name in this rail starts at the row's own 8px padding. That is the whole
rule, and it is why the tree finally reads as flush with the panel: an empty
leading slot reserved on every row (so that a folder's chevron would have a
column of its own) pushed all 139 names inward and made the column look
indented from nothing. Before that it was worse — five numbers reached a
nested name: 8px padding, a 16px slot, an 8px gap, a 14px nest margin, 6px
of nest padding and a 1px guide line, 51px in at 21px per level.
LEAD (+GAP) a glyph slot for the ONLY row kind that has a glyph: a
folder. Its name therefore sits one glyph in, and that offset
IS the hierarchy. Courses, chats and pager rows carry no
leading element, so their names sit at the panel edge.
INDENT the ONE nesting step, as a margin and nothing else. Folders
do not nest inside folders, so it is applied exactly once.
They are tokens because this list will grow row kinds (favourites,
recents), and a value written twice is the next misalignment. */
--ws-tree-indent: 12px;
/* One row height for BOTH tabs. A chat row and a course row are the same kind
of object — a single line you scan — so switching tabs must not reflow the
column. Depth is carried by whitespace alone (no rule, no guide line, no
card), and this is the whitespace. */
--ws-tree-row-h: 32px;
/* A leading GLYPH slot, for the rows that actually have a glyph: a folder.
Rows without one (courses, chats, pagers) do not reserve it — an empty slot
on every row is what pushed the whole tree right of the panel's own edge.
16px box + 8px of air, stated once. */
--ws-tree-lead: 16px;
--ws-tree-lead-gap: 8px;
/* THE height of a pane header, for every pane. It is a token rather than a
number in each rule because the two headers' bottom hairlines have to be
ONE line across the seam: the conversation's header and the classroom's
header are different objects (a title and a status word; a tab strip and a
button) and the moment either owns its own height they drift apart by a
pixel or two and the hairline visibly steps at the seam — which is exactly
what happened while the chat header sat at 44px and the classroom's at 42.
The value is the classroom's, because that one is DERIVED: 6px of strip
inset + a 31px tab + the 5px tail that joins the selected tab to the pane
below it. Change it here and both panes move together. */
--ws-pane-head-h: 42px;
/* Floating menus sit OVER the tree, so they are opaque where the rail is
translucent: a popover you can read the list through is not a popover. */
--ws-raised: #ffffff;
--ws-accent: #7c3aed; /* violet-600, the product's accent */
--ws-accent-thread: rgba(124, 58, 237, 0.32);
--ws-accent-wash: rgba(124, 58, 237, 0.08);
--ws-live: #10b981; /* emerald-500, as v2 */
--ws-fail: #f43f5e; /* rose-500, as v2 */
--ws-ease: cubic-bezier(0.22, 0.61, 0.36, 1);
/* Layered, not blurred. A contact shadow seats the card, then two wider and
softer casts carry it off the page. */
--ws-lift:
0 1px 1px rgba(15, 23, 42, 0.035), 0 3px 6px -2px rgba(15, 23, 42, 0.04),
0 10px 20px -8px rgba(15, 23, 42, 0.05), 0 26px 46px -22px rgba(15, 23, 42, 0.07);
/* Focus adds the violet thread the classic composer only hints at
(`focus-within:shadow-violet-500/[0.06]`), and commits to it. */
--ws-lift-focus:
0 0 0 1px var(--ws-accent-thread), 0 0 0 4px var(--ws-accent-wash),
0 2px 3px -1px rgba(15, 23, 42, 0.04), 0 8px 16px -4px rgba(15, 23, 42, 0.055),
0 22px 40px -14px rgba(15, 23, 42, 0.08), 0 40px 70px -34px rgba(124, 58, 237, 0.14);
--ws-lift-sm:
0 1px 1px rgba(15, 23, 42, 0.03), 0 2px 4px -1px rgba(15, 23, 42, 0.035),
0 6px 12px -6px rgba(15, 23, 42, 0.045);
--ws-lift-sm-hover:
0 1px 2px rgba(15, 23, 42, 0.04), 0 4px 8px -2px rgba(15, 23, 42, 0.055),
0 10px 18px -8px rgba(15, 23, 42, 0.06);
/* Grain. One inline SVG, tiled — no request, alpha baked in so it works over
any background. Tuned to be invisible on a bad panel. */
--ws-grain: url("data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' width='160' height='160'%3E%3Cfilter id='g'%3E%3CfeTurbulence type='fractalNoise' baseFrequency='0.85' numOctaves='3' stitchTiles='stitch'/%3E%3CfeColorMatrix type='saturate' values='0'/%3E%3C/filter%3E%3Crect width='160' height='160' filter='url(%23g)' opacity='0.04'/%3E%3C/svg%3E");
background: linear-gradient(180deg, var(--ws-canvas-top) 0%, var(--ws-canvas-bottom) 100%);
color: var(--ws-ink);
}
.dark .ws-root {
--ws-canvas-top: #020617; /* slate-950 */
--ws-canvas-bottom: #0f172a; /* slate-900 */
--ws-rail: rgba(15, 23, 42, 0.62);
--ws-surface: rgba(15, 23, 42, 0.78);
/* Dark's ramp was already the better of the two — it cleared AA at every step
that carries text — so it is lifted only at the top and the bottom, to keep
the ROLES the same distance apart in both themes:
ink 16.8:1 soft 11.7:1 mute 6.8:1 faint 3.6:1 (decoration)
Everything dark gains from this pass it gains from usage, not from the
ramp: counts and labels moved off `faint`, and the rail's buttons stopped
resting at 45% opacity. */
--ws-ink: #f8fafc; /* slate-50 */
--ws-ink-soft: #cbd5e1; /* slate-300 */
--ws-ink-mute: #94a3b8; /* slate-400 */
--ws-ink-faint: #64748b; /* slate-500 — decoration only */
--ws-line: rgba(255, 255, 255, 0.12);
--ws-line-soft: rgba(255, 255, 255, 0.07);
--ws-tint: rgba(255, 255, 255, 0.06);
--ws-tint-strong: rgba(255, 255, 255, 0.1);
--ws-raised: #1e293b; /* slate-800 — one step above the slate-900 canvas */
--ws-accent: #a78bfa; /* violet-400 reads on slate-950 */
--ws-accent-thread: rgba(167, 139, 250, 0.36);
--ws-accent-wash: rgba(167, 139, 250, 0.1);
--ws-lift:
0 1px 1px rgba(0, 0, 0, 0.3), 0 4px 10px -3px rgba(0, 0, 0, 0.32),
0 14px 28px -12px rgba(0, 0, 0, 0.38), 0 30px 54px -26px rgba(0, 0, 0, 0.44);
--ws-lift-focus:
0 0 0 1px var(--ws-accent-thread), 0 0 0 4px var(--ws-accent-wash),
0 2px 4px -1px rgba(0, 0, 0, 0.3), 0 10px 20px -5px rgba(0, 0, 0, 0.36),
0 26px 48px -16px rgba(0, 0, 0, 0.44), 0 44px 76px -36px rgba(167, 139, 250, 0.2);
--ws-lift-sm: 0 1px 2px rgba(0, 0, 0, 0.3), 0 4px 10px -4px rgba(0, 0, 0, 0.34);
--ws-lift-sm-hover: 0 2px 4px rgba(0, 0, 0, 0.34), 0 8px 16px -6px rgba(0, 0, 0, 0.4);
--ws-grain: url("data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' width='160' height='160'%3E%3Cfilter id='g'%3E%3CfeTurbulence type='fractalNoise' baseFrequency='0.85' numOctaves='3' stitchTiles='stitch'/%3E%3CfeColorMatrix type='saturate' values='0'/%3E%3C/filter%3E%3Crect width='160' height='160' filter='url(%23g)' opacity='0.085'/%3E%3C/svg%3E");
}
/* ── Atmosphere ──────────────────────────────────────────────────────────
The lamp rides on the scrolling column itself: `background-attachment` is
`scroll` by default, which pins a background to the element's own border
box, so it stays put while the content scrolls under it. No overlay layer,
no z-index ladder, nothing to intercept a click.
Its box is the FIRST SCREEN — one `100dvh` band, not a fixed 700px — and the
pool sits a little above that band's middle, where the lockup and composer
now are. The earlier cut lit the top 700px from `50% 1%`, which was right
while the hero was pinned to the top and wrong the moment it moved to the
centre: a light whose brightest point falls on empty space is a light
pointing at nothing. */
.ws-canvas {
background-image:
/* the pool behind hero + composer — violet-white, so the brand tints the
light rather than sitting in a swatch somewhere */
radial-gradient(58% 34% at 50% 43%, rgba(124, 58, 237, 0.062), rgba(124, 58, 237, 0) 72%),
/* a white core inside the pool, so the falloff is not one flat cone and
the composer sits in the brightest part of the canvas */
radial-gradient(40% 24% at 50% 47%, rgba(255, 255, 255, 0.95), rgba(255, 255, 255, 0) 74%),
var(--ws-grain);
background-repeat: no-repeat, no-repeat, repeat;
background-size:
100% 100dvh,
100% 100dvh,
160px 160px;
background-position:
top center,
top center,
top left;
}
.dark .ws-canvas {
background-image:
radial-gradient(62% 36% at 50% 43%, rgba(167, 139, 250, 0.075), rgba(167, 139, 250, 0) 70%),
radial-gradient(36% 22% at 50% 47%, rgba(226, 232, 240, 0.045), rgba(226, 232, 240, 0) 72%),
var(--ws-grain);
}
/* The rail is the product's chrome material — translucent white over the
canvas gradient, blurred, hairline edge. Same glass as the home's floating
toolbars, given a full-height panel to be. */
.ws-rail {
background-color: var(--ws-rail);
background-image: var(--ws-grain);
backdrop-filter: blur(14px) saturate(140%);
-webkit-backdrop-filter: blur(14px) saturate(140%);
box-shadow: inset -1px 0 0 0 var(--ws-line-soft);
}
/* ── Type ────────────────────────────────────────────────────────────────
Same Inter stack as the rest of the product; the change is that the hero is
now set at headline size with headline tracking instead of at body size. */
.ws-display {
font-size: clamp(26px, 2.7vw, 33px);
font-weight: 600;
line-height: 1.22;
letter-spacing: -0.021em;
color: var(--ws-ink);
-webkit-font-smoothing: antialiased;
text-rendering: optimizeLegibility;
}
.ws-subtitle {
font-size: 13.5px;
line-height: 1.55;
letter-spacing: 0.005em;
color: var(--ws-ink-mute);
}
/* Section labels. Wider tracking is what makes 10px read as a label rather
than as a very small heading — and 10px uppercase needs MORE contrast than
body text, not less, so the label sits at `soft` while its size and case do
the work of saying "this is a heading, not a row". */
.ws-label {
font-size: 10px;
font-weight: 600;
line-height: 1;
text-transform: uppercase;
letter-spacing: 0.145em;
color: var(--ws-ink-soft);
}
/* Counts are information — how many courses a folder holds — so they are read
at AA, not left at the 1.5:1 they used to sit at. */
.ws-count {
font-size: 10px;
font-weight: 500;
font-variant-numeric: tabular-nums;
letter-spacing: 0.05em;
color: var(--ws-ink-mute);
}
/* Hairlines that fade out, so panels dissolve into the canvas instead of
ending on a hard tick. */
.ws-seam-rail {
height: 1px;
border: 0;
background: linear-gradient(90deg, var(--ws-line) 0%, var(--ws-line) 58%, transparent 100%);
}
/* ── Rail rows ───────────────────────────────────────────────────────────
One soft neutral pill, filling the whole row, for hover / focus / selection.
That is the ENTIRE resting vocabulary: no left edge bar, no border, no
shadow, no separator. The violet bar that used to grow from the row's centre
line on hover and stay out on the selected row is gone — it was the loudest
mark in a column whose job is to be read past, and being the only thing at
the very left edge it also made the left edge look ragged. Selection is the
same pill one step stronger, which is a difference you can see without
introducing a second colour. */
.ws-row {
color: var(--ws-ink-soft);
border-radius: 8px;
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease),
box-shadow 160ms var(--ws-ease);
}
.ws-row:hover {
background-color: var(--ws-tint);
color: var(--ws-ink);
}
.ws-row:active {
background-color: var(--ws-tint-strong);
}
/* Focus is as visible as hover, and then some — same fill plus a ring. */
.ws-row:focus-visible {
outline: none;
background-color: var(--ws-tint);
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
/* Trailing metadata: a page count, a folder's size, a timestamp. Deliberately
the quietest type in the row — it is there to be glanced at, never read, and
at 11px in the metadata ink it was competing with the names on all 139 rows
at once.
`min-width` is the ⋯ button's own width: the action overlays this slot on
hover (see `.ws-row-wrap-actionable`), so reserving the same footprint means
the swap cannot change what the name has to work with. The margin stands in
for the flex gap the tree rows no longer carry — the leading slot is the
only gap in the row now, so the trailing one is stated here, once. */
.ws-row-meta {
color: var(--ws-ink-mute);
font-variant-numeric: tabular-nums;
font-size: 10px;
letter-spacing: 0.01em;
min-width: 24px;
margin-left: 8px;
text-align: right;
transition: opacity 160ms var(--ws-ease);
}
/* ── Course rows vs. chat rows ───────────────────────────────────────────
Both populations are compact, single-line lists. Course entries retain the
stronger 13px artifact name and put their page count at the far edge; chat
entries stay lighter and put time/status there. The surrounding tree vs.
stream structure supplies the remaining distinction without spending a
second line per course. */
.ws-course {
min-height: var(--ws-tree-row-h);
}
/* The darkest, heaviest text in the rail body — at REST, not on hover. This is
the thing the user came to the rail to find; making it wait for a pointer is
what made the whole column read as grey. */
.ws-course-name {
font-size: 13px;
font-weight: 550;
line-height: 1.25;
letter-spacing: -0.006em;
color: var(--ws-ink);
}
.ws-meta-sep {
color: var(--ws-ink-faint);
}
/* A folder heads its group: same ink as the courses under it, a shade heavier,
so the hierarchy reads without a second label style. */
.ws-folder-name {
font-size: 13px;
font-weight: 600;
letter-spacing: -0.004em;
color: var(--ws-ink);
}
/* Chats: one step back from the courses, in weight and in ink. Deliberately
NOT faint — a chat you can barely read is not "quiet", it is broken. */
.ws-chat-row {
/* No height of its own: a chat row is the same object as a course row, and
two numbers here is how the two tabs drifted apart. */
}
.ws-chat-name {
font-size: 13px;
font-weight: 450;
line-height: 1.2;
letter-spacing: 0.002em;
color: var(--ws-ink-soft);
}
.ws-row:hover .ws-chat-name,
.ws-row-active .ws-chat-name {
color: var(--ws-ink);
}
/* ── Rail actions ────────────────────────────────────────────────────── */
.ws-new {
background-color: var(--ws-surface);
border: 1px solid var(--ws-line);
color: var(--ws-ink-soft);
box-shadow: var(--ws-lift-sm);
letter-spacing: 0.005em;
transition:
border-color 180ms var(--ws-ease),
color 180ms var(--ws-ease),
box-shadow 180ms var(--ws-ease),
transform 180ms var(--ws-ease);
}
.ws-new:hover {
border-color: rgba(15, 23, 42, 0.16);
color: var(--ws-ink);
transform: translateY(-1px);
box-shadow: var(--ws-lift-sm-hover);
}
.dark .ws-new:hover {
border-color: rgba(255, 255, 255, 0.18);
}
.ws-new:focus-visible {
outline: none;
border-color: var(--ws-accent-thread);
box-shadow: 0 0 0 3px var(--ws-accent-wash);
}
.ws-quiet {
color: var(--ws-ink-mute);
transition: color 160ms var(--ws-ease);
}
.ws-quiet:hover {
color: var(--ws-ink);
}
.ws-skeleton {
background-color: var(--ws-tint-strong);
}
/* ── Status dots ─────────────────────────────────────────────────────────
Same emerald/rose/neutral the v2 shell used; a running session gets a
small spinner instead of the old breathing halo — live states spin, static
states stay put. Every session row reserves the SAME slot for its mark, so
the 12px spinner and the 6px dot hold identical layout width and the names
in the list line up column-to-column whichever state a row is in. */
.ws-status-slot {
/* A TRAILING mark now, beside the timestamp — not a leading one. A chat's
status is metadata like its time, and putting it at the head of the row was
what made every chat name start a slot right of every course name. The box
stays fixed so the 12px spinner and the 6px dot hold identical width and a
running row does not shift its neighbours. */
width: 16px;
height: 16px;
margin-left: 6px;
flex-shrink: 0;
display: grid;
place-items: center;
}
.ws-dot {
width: 6px;
height: 6px;
border-radius: 999px;
flex-shrink: 0;
/* A 6px graphic still has to be seen: `faint` is for rules, not for the mark
that tells you a chat finished. */
background: var(--ws-ink-mute);
}
.ws-spin-live {
width: 12px;
height: 12px;
flex-shrink: 0;
color: var(--ws-live);
animation: ws-spin 0.9s linear infinite;
}
.ws-dot-fail {
background: var(--ws-fail);
}
@keyframes ws-spin {
to {
transform: rotate(360deg);
}
}
/* ── The composer ────────────────────────────────────────────────────────
The classic face's chrome, given a focused state worth having. Applied
through ProLaunchPanel's existing `workspace` variant, so `/` is untouched. */
.ws-composer {
background-color: var(--ws-surface);
backdrop-filter: blur(16px) saturate(150%);
-webkit-backdrop-filter: blur(16px) saturate(150%);
border: 1px solid var(--ws-line);
border-radius: 16px;
box-shadow: var(--ws-lift);
transition:
border-color 220ms var(--ws-ease),
box-shadow 260ms var(--ws-ease),
transform 260ms var(--ws-ease);
}
.ws-composer:hover {
border-color: rgba(15, 23, 42, 0.15);
}
.dark .ws-composer:hover {
border-color: rgba(255, 255, 255, 0.17);
}
.ws-composer:focus-within {
border-color: transparent;
box-shadow: var(--ws-lift-focus);
transform: translateY(-1px);
}
.ws-composer textarea::placeholder {
color: var(--ws-ink-mute);
opacity: 0.75;
}
/* ── Entrance ────────────────────────────────────────────────────────────
One orchestrated load, four beats. `both` holds the from-state through the
delay, which is why the reduced-motion block has to hand opacity back. */
@keyframes ws-rise {
from {
opacity: 0;
transform: translateY(10px);
}
to {
opacity: 1;
transform: none;
}
}
@keyframes ws-fade {
from {
opacity: 0;
}
to {
opacity: 1;
}
}
.ws-enter {
animation: ws-rise 620ms var(--ws-ease) both;
}
.ws-enter-rail {
animation: ws-fade 500ms var(--ws-ease) both;
}
.ws-d1 {
animation-delay: 40ms;
}
.ws-d2 {
animation-delay: 150ms;
}
.ws-d3 {
animation-delay: 260ms;
}
.ws-d4 {
animation-delay: 380ms;
}
@media (prefers-reduced-motion: reduce) {
.ws-enter,
.ws-enter-rail {
animation: none;
opacity: 1;
transform: none;
}
.ws-spin-live {
animation: none;
}
.ws-composer:focus-within,
.ws-new:hover {
transform: none;
}
}
/* ── The opening block ──────────────────────────────────────────────────
Lockup and composer, and then the feed right under them. Not a screen: a
block that ends where its content ends.
Two earlier rounds are buried in this rule. The first packed the whole feed
into one viewport and made the front door read as a page with a search box
on it. The second gave the group a full viewport of its own — `min-height:
100dvh` and `justify-content: center` — which fixed the hero and created a
worse problem underneath it: ~250px of reserved nothing between the composer
and the section head, there for no reason except that the box had promised
to be a screen tall.
So the group keeps the POSITION that centring gave it and loses the box that
reserved the space. `48vh - 127px` is not a magic number: it is what
`(100dvh - lift - content)/2` evaluates to for this content, algebraically,
across every window between roughly 600px and 1100px tall. Stating it as a
top offset rather than as centring buys two things — the block stops
reserving height it does not use, and a composer that grows while you type
pushes DOWN instead of dragging the lockup up the screen.
`max()` is the short-window floor: under about 320px of window the formula
goes negative, and the lockup must not be pushed off the top of the scroll
origin. Below `md` the compact exit bar takes 48px off the column, and
`--ws-home-chrome` gives it back so the group does not sit lower there. */
.ws-home-first {
--ws-home-pad: clamp(24px, 4vh, 48px);
/* Ordinary section spacing, in tens of pixels. This is the whole gap between
the composer and the discover head now — the feed follows the composer
rather than waiting a screen for it. */
--ws-home-gap: clamp(48px, 7vh, 88px);
display: flex;
flex-direction: column;
padding-top: max(var(--ws-home-pad), calc(48vh - 127px - var(--ws-home-chrome, 0px) / 2));
padding-bottom: var(--ws-home-gap);
}
@media (max-width: 767px) {
.ws-home-first {
--ws-home-chrome: 48px;
}
}
/* ── Featured seam ───────────────────────────────────────────────────────
Nothing to draw and nothing to correct. The feed starts one ordinary section
gap under the composer (`--ws-home-gap` above), so the head is on screen
without scrolling and says for itself what is below it. The chevron that
used to point across the old gap went with the gap: an arrow aimed at
content the reader can already see is a caption on a photograph of itself.
What is left here is the positioning context. */
.ws-featured {
position: relative;
}
/* ── The hero lockup ─────────────────────────────────────────────────────
The workspace's identity mark, not its billboard. `logo-horizontal.png` is
1232×269, so an unsized `<img>` renders at 269px tall and turns the page
into a splash screen — the height below is what keeps the composer the
protagonist. Fixed pixels, deliberately: a viewport-relative hero grows on
a wide monitor, which is exactly where a 760px column can least afford it.
For scale: the rail's wordmark is 21px, the classic homepage hero is 48–64px
(a full page with nothing under it). The first cut of this surface took the
bottom of that range and went under it — 40px is a browser-tab favicon, and
a page whose masthead weighs nothing is top-light no matter what sits below.
46/56px is the correction: at 4.58:1 the wordmark lands 256px wide against a
760px composer, about a third of it, which is the proportion that makes the
two read as one lockup instead of a caption over a box. */
.ws-hero-logo {
height: 46px;
width: auto;
max-width: 100%;
user-select: none;
-webkit-user-drag: none;
}
@media (min-width: 768px) {
.ws-hero-logo {
height: 56px;
}
}
/* One line under the mark, at the ink ramp's quiet step — a subtitle, not a
second headline. It steps up with the mark: 13px under a 56px wordmark reads
as fine print rather than as the lockup's own line. */
.ws-tagline {
font-size: 14px;
line-height: 1.5;
color: var(--ws-ink-mute);
}
/* ── The hero mark ───────────────────────────────────────────────────────
The wordmark already sits in the rail, so the hero takes the mark alone:
the page gets the weight it was missing without saying the product's name
twice. Modest — roughly two line-heights of the headline — and lit from
below by the same violet the canvas pool uses, so it belongs to the light
rather than sitting on top of it. */
.ws-hero-mark {
width: 46px;
height: 46px;
border-radius: 12px;
filter: drop-shadow(0 6px 14px rgba(124, 58, 237, 0.22))
drop-shadow(0 1px 2px rgba(15, 23, 42, 0.1));
user-select: none;
-webkit-user-drag: none;
}
.dark .ws-hero-mark {
filter: drop-shadow(0 6px 18px rgba(167, 139, 250, 0.26))
drop-shadow(0 1px 2px rgba(0, 0, 0, 0.4));
}
/* ── The tab switch ──────────────────────────────────────────────────────
Chat | courses, over one body — a SEGMENTED CONTROL of two equal halves.
It used to be an underline strip sized to its labels and parked at the left,
because a filled segment "reads as a mode" while tabs read as views. Two
lists that take turns owning the whole body ARE a mode, and at rail width
the underline was the quietest mark in a column of louder ones: the selected
tab took a 2px hairline while every row below it could take a filled pill.
So the switch spends the width it has — one tinted track, two equal halves,
the selected half lifted onto a solid ground with a hairline and a contact
shadow. Same materials as the compose button above it (surface + line +
`lift-sm`), one notch of the accent kept for the selected count, so the
strip is louder without introducing a colour the rail does not already own.
The track is `mx-3`, flush with the compose button and the findrow, so the
three stack as one control column.
Equal halves cost the wider label its last characters — that was the whole
objection to them — and the fix is TYPE, not width (see `.ws-navtab-label`
below), which buys back roughly a third of the label's box. */
.ws-navtabs {
height: 34px;
padding: 3px;
gap: 3px;
border-radius: 10px;
background-color: var(--ws-tint);
box-shadow: inset 0 0 0 1px var(--ws-line-soft);
}
/* Two halves, always: `flex: 1 1 0` (not `auto`) so the wider label cannot
claim more of the track than the narrower one, and `min-width: 0` so it
truncates inside its own half instead of pushing the track open. */
.ws-navtab {
position: relative;
flex: 1 1 0;
min-width: 0;
height: 100%;
padding: 0 8px;
border-radius: 7px;
color: var(--ws-ink-mute);
transition:
color 160ms var(--ws-ease),
background-color 160ms var(--ws-ease),
box-shadow 160ms var(--ws-ease);
}
/* The unselected half stays quiet — it is the other view, not a second
destination competing with the one you are reading. */
.ws-navtab:hover:not([aria-selected='true']) {
color: var(--ws-ink-soft);
background-color: var(--ws-tint);
}
.ws-navtab:focus-visible {
outline: none;
color: var(--ws-ink);
box-shadow:
0 0 0 1px var(--ws-accent-thread),
0 0 0 3px var(--ws-accent-wash);
}
/* The selected half. A solid ground rather than a mark drawn under the label:
at 200px a 2px underline is the quietest mark in a column of louder ones. */
.ws-navtab[aria-selected='true'] {
color: var(--ws-ink);
background-color: rgba(255, 255, 255, 0.92);
box-shadow:
inset 0 0 0 1px var(--ws-line-soft),
var(--ws-lift-sm);
}
/* Dark inverts which way "raised" points: a chip that is DARKER than its track
reads as a hole, so the selected half goes up the white ramp instead, and
drops the cast shadow (a shadow on slate-900 is noise, not lift). */
.dark .ws-navtab[aria-selected='true'] {
background-color: rgba(255, 255, 255, 0.12);
box-shadow: inset 0 0 0 1px var(--ws-line);
}
.ws-navtab[aria-selected='true'] .ws-navtab-label {
font-weight: 600;
}
/* Sentence case at 11px, no tracking — NOT the rail's 10px uppercase section
label. That label was built for a full-width heading with the whole rail to
spread over; inside half of a 200px track, 0.145em of tracking on seven
uppercase characters is the difference between "Courses" and "COU…".
It stays a label by size and by ink, not by case. */
.ws-navtab-label {
font-size: 11px;
font-weight: 500;
line-height: 1;
letter-spacing: 0.005em;
color: inherit;
}
/* ── Find, under the tabs ────────────────────────────────────────────────
The prototype's findrow, copied in form and density: one always-visible
icon+input row heading the tab content, filtering whichever list the
active tab shows. 30px, tint ground, radius 8 — and the focus state lifts
it onto the panel surface with the violet thread and wash, the same
sentence the composer speaks. */
.ws-find {
display: flex;
align-items: center;
gap: 7px;
height: 30px;
padding: 0 9px;
border-radius: 8px;
background-color: var(--ws-tint);
border: 1px solid transparent;
color: var(--ws-ink-mute);
transition:
background-color 160ms var(--ws-ease),
border-color 160ms var(--ws-ease),
box-shadow 160ms var(--ws-ease);
}
.ws-find:focus-within {
background-color: var(--ws-surface);
border-color: var(--ws-accent-thread);
box-shadow: 0 0 0 3px var(--ws-accent-wash);
}
.ws-find-input {
border: 0;
background: none;
outline: none;
width: 100%;
font-size: 13px;
color: var(--ws-ink);
}
.ws-find-input::placeholder {
color: var(--ws-ink-mute);
}
/* The header row's icon button — the new-folder button on the courses tab. The
find input's own height, so the two read as one control group. */
.ws-find-icon {
display: inline-flex;
align-items: center;
justify-content: center;
flex-shrink: 0;
width: 30px;
height: 30px;
border-radius: 8px;
color: var(--ws-ink-mute);
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease);
}
.ws-find-icon:hover {
background-color: var(--ws-tint);
color: var(--ws-ink);
}
.ws-find-icon:focus-visible {
outline: none;
color: var(--ws-ink);
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
/* Folder creation and renaming are one row in the tree, with the input starting
in the same column as the names around it (the row reserves the rail's
leading slot). Focus stays inside that row instead of promoting it into a
detached input card. */
.ws-inline-folder {
color: var(--ws-ink-mute);
transition:
background-color 160ms var(--ws-ease),
box-shadow 160ms var(--ws-ease);
}
.ws-inline-folder:focus-within {
background-color: var(--ws-tint);
box-shadow: inset 0 0 0 1px var(--ws-accent-thread);
}
.ws-inline-folder-input {
width: 100%;
border: 0;
background: none;
outline: none;
font-size: 12.5px;
color: var(--ws-ink);
caret-color: var(--ws-accent);
}
.ws-inline-folder-input::placeholder {
color: var(--ws-ink-mute);
}
.ws-inline-folder-input:disabled {
cursor: wait;
}
.ws-inline-folder-action {
display: inline-flex;
width: 22px;
height: 22px;
flex-shrink: 0;
/* The tree rows carry no flex gap (the leading slot is the row's only one),
so the confirm/cancel pair states its own spacing here. */
margin-left: 4px;
align-items: center;
justify-content: center;
border-radius: 6px;
color: var(--ws-ink-mute);
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease),
opacity 160ms var(--ws-ease);
}
.ws-inline-folder-action:hover:not(:disabled) {
background-color: var(--ws-tint);
color: var(--ws-ink);
}
.ws-inline-folder-action:focus-visible {
outline: none;
color: var(--ws-ink);
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
.ws-inline-folder-action:disabled {
cursor: not-allowed;
opacity: 0.38;
}
.ws-inline-folder-error {
/* Under the input it belongs to, starting in the same column as the folder
name above it: `0.5rem` is the row's own `px-2`, then the leading glyph slot
and its air. */
padding-left: calc(0.5rem + var(--ws-tree-lead) + var(--ws-tree-lead-gap));
color: var(--ws-fail);
font-size: 10.5px;
line-height: 14px;
}
/* Folder glyphs: present, never loud. */
.ws-glyph {
color: var(--ws-ink-mute);
}
/* ── The classroom picker ────────────────────────────────────────────────── */
.ws-cmenu-row {
height: var(--ws-cmenu-row-h);
}
/* One flat list, so the window is exactly N rows tall — no sticky section
labels eating a row, and no half-row at rest. */
.ws-cmenu-scroll {
max-height: calc(var(--ws-cmenu-row-h) * var(--ws-cmenu-rows));
scrollbar-width: thin;
}
.ws-cmenu-scroll::-webkit-scrollbar {
width: 6px;
}
.ws-cmenu-scroll::-webkit-scrollbar-thumb {
border-radius: 3px;
background: var(--ws-line);
}
.ws-cmenu-scroll::-webkit-scrollbar-track {
background: transparent;
}
/* The leading glyph slot — only on rows that carry a glyph (a folder, a folder
being created). Courses, chats and pager rows have no leading element at all,
so their names begin at the row's own `px-2`: the panel edge, which is where
the eye already is. A folder's name sits one glyph further in, and that offset
IS the hierarchy — it is not a misalignment to be padded away on every other
row. */
.ws-tree-lead {
display: grid;
place-items: center;
flex: 0 0 var(--ws-tree-lead);
width: var(--ws-tree-lead);
margin-right: var(--ws-tree-lead-gap);
color: var(--ws-ink-mute);
}
/* Every row in the tree, folders and courses alike. */
.ws-tree-row {
height: var(--ws-tree-row-h);
}
/* Nested rows sit ONE step in, and the step is a margin — nothing else. There
used to be a guide line here (a `border-left`, plus the 6px of padding that
held it off the names), which is exactly the kind of vertical rule this tree
does not use: depth is one modest inset under the folder row that owns it. */
.ws-nest {
margin-left: var(--ws-tree-indent);
}
/* The tree's top level holds two kinds of row: folders, and the courses that
are in no folder. The loose run takes no indent — an inset would put it at a
folder member's depth and read as membership in the folder above it — so the
only thing marking the change of kind is space, and enough of it to read as a
break rather than a nudge. Padding, not a margin: this div is a child of a
`space-y-px` list, whose own margin rule outranks a single class. */
.ws-loose {
padding-top: 10px;
}
/* Show-more / collapse — a plain line of quiet text belonging to the list it
extends, NOT a control with a glyph on it. It used to carry a chevron, which
made a pager row look like a third kind of disclosure in a tree that already
has one, and put its label in a column of its own. It reserves the rail's
leading slot instead, so the label starts exactly where the names above it
do. */
.ws-more {
color: var(--ws-ink-mute);
font-weight: 500;
border-radius: 8px;
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease);
}
.ws-more:hover {
background-color: var(--ws-tint);
color: var(--ws-ink);
}
.ws-more:focus-visible {
outline: none;
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
/* ── Row + its hover-revealed action ─────────────────────────────────────
The action is a SIBLING of the row button (no nested buttons), so the wrap
carries the hover state for both. */
.ws-row-wrap:hover .ws-row,
.ws-row-wrap:focus-within .ws-row {
background-color: var(--ws-tint);
color: var(--ws-ink);
}
.ws-row-wrap {
transform-origin: center;
transition:
transform 160ms var(--ws-ease),
box-shadow 160ms var(--ws-ease);
}
/* The action takes the metadata's slot rather than pushing it: re-truncating
the name on hover would make every row twitch as the pointer crossed it. The
two occupy the same reserved footprint (`.ws-row-meta`'s `min-width` is this
button's width) and the action is out of flow entirely, so the name's usable
width is identical at rest and on hover.
Scoped to rows that HAVE an action — a chat row in the read-only drawer has
none, so its metadata would otherwise vanish on hover for nothing. */
.ws-row-wrap-actionable:hover .ws-row-meta,
.ws-row-wrap-actionable:focus-within .ws-row-meta {
opacity: 0;
}
.ws-row-trailing {
opacity: 0;
transition: opacity 160ms var(--ws-ease);
}
.ws-row-wrap:hover .ws-row-trailing,
.ws-row-trailing:focus-within {
opacity: 1;
}
/* Revealed on hover — a per-row ⋯ shown on all 139 rows at once would be a
column of noise — but once revealed it has to look pressable, which a bare
grey glyph did not. It gets the rail's own surface and a hairline. */
.ws-row-action {
color: var(--ws-ink-mute);
background-color: var(--ws-raised);
box-shadow: 0 0 0 1px var(--ws-line);
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease);
}
.ws-row-action:hover {
background-color: var(--ws-raised);
color: var(--ws-ink);
box-shadow:
0 0 0 1px var(--ws-line),
var(--ws-lift-sm);
}
.ws-row-action:focus-visible {
outline: none;
color: var(--ws-ink);
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
/* ── Rail foot ───────────────────────────────────────────────────────────
SiteHeader's own widgets, dropped into the rail. They bring their own
(grey-ramp) styling; this only quiets them to the rail's contrast and gives
the cluster a divider of the rail's own kind. */
.ws-utils {
color: var(--ws-ink-mute);
}
.ws-utils-sep {
width: 1px;
height: 16px;
margin: 0 4px;
background: var(--ws-line);
}
/* ── Resize handle ───────────────────────────────────────────────────────
A 6px hit strip straddles each pane edge, but the only visible affordance is
the seam itself: a fine resting thread that strengthens on hover. On a seam
that can collapse from its seam — the navigation rail, and only it — the same
hover reveals its sibling grip. Everything else folds from its own pane
header (`PaneFoldButton`): a seam between two collapsible panes can only ever
speak for one of them, and carrying both put two reversed chevrons at one
height next to a third. */
.ws-resize {
touch-action: none;
}
.ws-resize-thread {
position: absolute;
inset-block: 0;
left: 50%;
width: 1px;
transform: translateX(-50%);
background: var(--ws-line);
transition:
width 140ms var(--ws-ease),
background-color 140ms var(--ws-ease),
box-shadow 160ms var(--ws-ease);
}
.ws-resize:hover .ws-resize-thread,
.ws-resize:focus-within .ws-resize-thread {
width: 2px;
background: var(--ws-ink-faint);
}
/* In flight. `data-ws-drag` is set on the handle being dragged, not on the
document — the document flag below is shared by every seam on the page, and
lighting all of them up while one is moving is a lie about what is
happening. Released, both attributes go and the seam settles back. */
.ws-resize[data-ws-drag='true'] .ws-resize-thread {
width: 2px;
background: var(--ws-accent);
box-shadow: 0 0 0 3px var(--ws-accent-wash);
}
/* While dragging, the cursor must not flicker as the pointer crosses elements
and nothing may select under it — both are set on the document, because the
pointer is captured and can travel far outside the handle. */
[data-ws-resizing] {
cursor: col-resize !important;
user-select: none !important;
}
/* A narrow workspace shows one content pane at a time. When the classroom is
the visible pane, its chat-return tab lands on the navigation edge: that is
the nav/classroom seam, not a second strip beside it. Centre the existing
right-pointing return control on the thread; clicking it switches focus to
chat, which is the narrow-layout form of folding the classroom on the right.
Nothing has to be suppressed alongside it any more — the rail's fold moved
into its own header, so the seam carries only the resize thread. */
.ws-root:has(> .ws-tab:not(.ws-pane-last)) > .ws-tab:not(.ws-pane-last) {
z-index: 30;
margin-inline: -15px;
background-color: transparent;
background-image: none;
box-shadow: none;
}
.ws-root:has(> .ws-tab:not(.ws-pane-last)) > .ws-tab:not(.ws-pane-last):hover {
background-color: transparent;
}
.ws-root:has(> .ws-tab:not(.ws-pane-last):hover) > .ws-rail .ws-resize-thread,
.ws-root:has(> .ws-tab:not(.ws-pane-last):focus-visible) > .ws-rail .ws-resize-thread {
width: 2px;
background: var(--ws-ink-faint);
}
/* No transition on the rail's width: it is driven straight from the pointer,
and easing a value that already tracks the cursor only adds lag. */
@media (prefers-reduced-motion: reduce) {
.ws-row-wrap,
.ws-row-trailing,
.ws-navtab::after {
transition: none;
}
}
/* The open course / the attached conversation: the SAME neutral pill hover
draws, one step stronger and permanent. No accent bar, no border, no second
colour — "what am I looking at" is answered by one fill that covers the whole
row, which is also why selection cannot shift a single pixel of the text. */
.ws-row-active {
color: var(--ws-ink);
background: var(--ws-tint-strong);
}
/* ── Panes ───────────────────────────────────────────────────────────────
The three-pane surface: navigation, conversation, classroom. Each is an
independent column with its own header and its own collapse, and each is
made of the SAME materials the rail already uses — glass over the canvas
gradient, a hairline seam, the slate ramp. Nothing here is a new visual
idea; the pane is a rail that happens to be wide.
The seam is `inset box-shadow` rather than `border`, for the reason the
rail's is: a border participates in layout, so a resizing column would
twitch by a pixel as it is dragged. */
.ws-pane {
position: relative;
min-height: 0;
background-color: var(--ws-surface);
background-image: var(--ws-grain);
box-shadow: inset -1px 0 0 0 var(--ws-line-soft);
}
/* The last column carries no right seam — it ends at the window. */
.ws-pane-last {
box-shadow: none;
}
/* Pane headers. One row, one height for every pane (`--ws-pane-head-h`), the
same label/ink ramp as a rail section head: the panes are chrome, and chrome
that competes with the content it frames is the thing this pass keeps
removing.
The height lives ONLY here. A pane that sets its own would break the one
thing this row has to get right — the bottom hairline running unbroken across
the seam, with both fold chevrons on one baseline. */
.ws-pane-head {
height: var(--ws-pane-head-h);
background: linear-gradient(180deg, rgba(255, 255, 255, 0.4), rgba(255, 255, 255, 0));
box-shadow: inset 0 -1px 0 0 var(--ws-line-soft);
}
.dark .ws-pane-head {
background: linear-gradient(180deg, rgba(255, 255, 255, 0.035), rgba(255, 255, 255, 0));
}
/* The classroom's controls share one compact browser-chrome row. Its body then
starts on the 4px spacing scale instead of inheriting a roomy studio top
inset on top of the pane header.
The studio frame keeps `ContainBox`'s default centre-centre fit, so the
16:9 slide and the pager that rides its lower edge hang together as one
group in the middle of the pane. Top-aligning the host here was a dead-band
workaround (975032d594) whose real culprit — a redundant header — is gone:
with the slide top-aligned and the pager frame-anchored, a tall pane put
the slide at the top and the pager at the bottom, scattering one page's
chrome across the pane. The 12px frame padding stays the only gap between
pane chrome and content. */
.ws-classroom-head {
gap: 6px;
}
.ws-classroom-body [data-maic-studio-frame='true'] {
padding-top: 12px;
}
/* SlideNavRail inside the workbench panel drops its standalone-classroom
header (`h-10 + mt-3 + mb-1`) entirely — the header's only panel child was
the collapse control, which now lives as a chevron on the rail/slide
boundary (see SlideNavRail.tsx) — so the first thumbnail starts on the same
12px rhythm as the canvas with no overlay positioning needed. The chevron
is styled in the component: a bare faint 12px mark at the rail's right
edge, vertically centred, deliberately lighter than the pane-seam grip
above (no pill, no border, no boxed glyph — a whisper of ground on hover
only), because it marks a boundary INSIDE a pane, not a pane seam. */
.ws-pane-eyebrow {
font-size: 9.5px;
font-weight: 600;
line-height: 1;
text-transform: uppercase;
letter-spacing: 0.14em;
color: var(--ws-ink-mute);
}
.ws-pane-title {
font-size: 12.5px;
font-weight: 500;
line-height: 1.3;
color: var(--ws-ink);
}
.ws-pane-sub {
font-weight: 400;
color: var(--ws-ink-mute);
}
/* The read-only tell on a saved course. A chip, not a banner: the absence of
edit affordances is the real signal — this only names it. */
.ws-readonly {
display: inline-flex;
align-items: center;
gap: 4px;
height: 20px;
padding: 0 7px;
border-radius: 999px;
font-size: 10.5px;
letter-spacing: 0.01em;
color: var(--ws-ink-soft);
background: var(--ws-tint);
box-shadow: inset 0 0 0 1px var(--ws-line-soft);
}
/* ── The right pane's tab strip ───────────────────────────────────────────
The header's name half, turned into a tablist. Multiple open courses cost no
new row of chrome: a persistent shelf was refused in the rail, and a second
row here would be the same mistake in a different pane. A hairline marks
where scope changes — everything LEFT of it acts on one tab, everything
RIGHT of it acts on the whole pane. That is the whole answer to the
×-means-two-things problem browsers never solved.
SHRINK TO A FLOOR, THEN SCROLL. There is no cap on tabs, so the strip has to
degrade gracefully forever: a `min-width` floor below which a tab will not
compress, and `overflow-x` past it. The drawn scrollbar lives in the
header's existing height, so the header does not grow a row for it. The
native scrollbar is an overlay scrollbar here — 0px reserved, nothing painted until you are
already scrolling — hence suppressed and redrawn. */
.ws-ctabs {
display: flex;
align-items: flex-start;
align-self: stretch;
gap: 0;
flex: 1;
min-width: 0;
height: auto;
padding-top: 6px;
overflow-x: auto;
overflow-y: hidden;
overscroll-behavior-x: contain;
scroll-behavior: smooth;
/* The fade is 26px wide, so scrolling a tab "into view" has to clear 26px or
it lands under the fade and arrives half-erased. */
scroll-padding-inline: 26px;
scrollbar-width: none;
background-repeat: no-repeat, no-repeat;
-webkit-mask-repeat: no-repeat;
mask-repeat: no-repeat;
}
.ws-ctabs::-webkit-scrollbar {
display: none;
}
/* The drawn scrollbar: thumb layer over track layer. Only the size and the x
offset are set from JS; everything else stays here. The thumb is on the
`faint` step rather than a tint — it is 3px of decoration whose entire job is
to be NOTICED, and at a tint's 9% it measured invisible on screen. */
.ws-ctabs:not([data-more='none']) {
background-image:
linear-gradient(var(--ws-ink-faint), var(--ws-ink-faint)),
linear-gradient(var(--ws-line), var(--ws-line));
}
/* A hard clip reads as damage; a fade reads as continuation. */
.ws-ctabs[data-more='right'] {
-webkit-mask-image: linear-gradient(90deg, #000 calc(100% - 26px), transparent);
mask-image: linear-gradient(90deg, #000 calc(100% - 26px), transparent);
}
.ws-ctabs[data-more='left'] {
-webkit-mask-image: linear-gradient(270deg, #000 calc(100% - 26px), transparent);
mask-image: linear-gradient(270deg, #000 calc(100% - 26px), transparent);
}
.ws-ctabs[data-more='both'] {
-webkit-mask-image: linear-gradient(
90deg,
transparent,
#000 26px,
#000 calc(100% - 26px),
transparent
);
mask-image: linear-gradient(90deg, transparent, #000 26px, #000 calc(100% - 26px), transparent);
}
/* The floor is 86px, and it is derived, not guessed: 15px of padding leaves
71px of name, which is five CJK glyphs at 13px — four once the run dot has
taken its 12px. At 70px the name gets 32px, i.e. one character and an
apology. */
.ws-ctab {
position: relative;
z-index: 0;
display: flex;
align-items: center;
gap: 6px;
flex: 0 1 auto;
min-width: 86px;
max-width: 250px;
height: 31px;
padding: 0 6px 0 9px;
border-radius: 7px 7px 0 0;
font-size: 13px;
font-weight: 500;
color: var(--ws-ink-mute);
background: transparent;
cursor: pointer;
transition:
background-color 140ms var(--ws-ease),
color 140ms var(--ws-ease);
}
/* Browser tabs communicate selection through topology, not a louder chip. The
selected surface rises to the header edge and its short tail covers the
header hairline, visually joining the classroom below. It still gets more of
the row than the tabs you are not reading.
Its floor is 113px, not 86px: its × is permanently visible and permanently in
the name's padding, so 86 + 27 is what leaves it the same five glyphs its
neighbours get. At a uniform 86px every tab showed five characters except the
one you were reading, which showed two — the active tab must never identify
itself less well than the tabs beside it. */
.ws-ctab-on {
flex: 1.6 1 auto;
min-width: 113px;
z-index: 1;
background: var(--ws-raised);
color: var(--ws-accent);
font-weight: 600;
box-shadow:
inset 1px 0 0 var(--ws-line),
inset -1px 0 0 var(--ws-line),
inset 0 1px 0 var(--ws-line),
0 -2px 7px -5px var(--ws-ink-faint);
}
.ws-ctab-on::after {
position: absolute;
right: 0;
bottom: -5px;
left: 0;
height: 5px;
content: '';
background: var(--ws-raised);
box-shadow:
inset 1px 0 0 var(--ws-line),
inset -1px 0 0 var(--ws-line);
}
/* Only background tabs divide from one another. The active surface already
supplies a stronger, structural boundary at either edge. */
.ws-ctab:not(.ws-ctab-on) + .ws-ctab:not(.ws-ctab-on)::before {
position: absolute;
top: 6px;
bottom: 6px;
left: 0;
width: 1px;
content: '';
background: var(--ws-line);
}
.ws-ctab:not(.ws-ctab-on):hover {
background: var(--ws-tint-strong);
color: var(--ws-ink-soft);
}
.ws-ctab:focus-visible {
outline: 2px solid var(--ws-accent);
outline-offset: -2px;
}
.ws-ctab-name {
flex: 1;
min-width: 0;
overflow: hidden;
text-overflow: ellipsis;
white-space: nowrap;
text-align: left;
}
/* Per-tab close. It is OUT OF FLOW, which is a stronger guarantee than
reserving space for it: out-of-flow cannot affect any width, so the invariant
"revealing × never moves a tab" holds by construction rather than by
discipline, and every tab spends those 23px on its name instead.
A PERMANENT CONTROL GETS A PERMANENT SLOT; A TRANSIENT ONE BORROWS. On the
selected tab the × is always visible, so the name reserves 23px of padding
and text never runs beneath it. On the others it appears only under hover or
focus and borrows the tail of the name for as long as it is there.
Selectedness does not change on hover, so neither geometry does.
THE × MUST NOT SHOUT. It is an exit affordance, not a label — on the
selected tab, where it is permanently visible, it rests at MUTED ink pulled
down to 60% opacity with no chip, so the eye lands on the course name, not
on the way out. Hovering the tab, either tab, is what arms it: soft ink at
90% on a raised chip. Pointing directly at the × goes to full ink at full
opacity. One register for every tab, active or not — the accent belongs to
the tab's identity, never to its close glyph. */
.ws-ctab-x {
position: absolute;
z-index: 2;
right: 4px;
top: 50%;
transform: translateY(-50%);
display: grid;
place-items: center;
width: 22px;
height: 22px;
border-radius: 6px;
color: inherit;
opacity: 0;
pointer-events: none;
transition:
background-color 130ms var(--ws-ease),
color 130ms var(--ws-ease),
opacity 130ms var(--ws-ease);
}
.ws-ctab-on .ws-ctab-name {
padding-right: 27px;
}
.ws-ctab-on .ws-ctab-x {
color: var(--ws-ink-mute);
opacity: 0.6;
pointer-events: auto;
}
/* Because it can sit on top of the name, a visible × carries its own chip so it
never reads as a glyph in the title. The chip has to read RAISED in both
themes: white is a step up from the light tab's tint, but in dark mode the
panel colour is DARKER than the tab's hover tint and reads recessed, so dark
mode lifts with light instead of painting a surface. */
.ws-ctab:hover .ws-ctab-x,
.ws-ctab:focus-within .ws-ctab-x {
color: var(--ws-ink-soft);
opacity: 0.9;
pointer-events: auto;
background: var(--ws-raised);
box-shadow: 0 0 0 1px var(--ws-line-soft);
}
.dark .ws-ctab:hover .ws-ctab-x,
.dark .ws-ctab:focus-within .ws-ctab-x {
background: rgba(255, 255, 255, 0.16);
box-shadow: 0 0 0 1px rgba(255, 255, 255, 0.14);
}
/* Directly over the × it goes to full ink at full opacity — still on a chip.
The scope is repeated so this outranks the armed state above (4 classes to
3): pointing at the × is the one action that must read as unmistakably
armed. */
.ws-ctab:hover .ws-ctab-x:hover,
.ws-ctab:focus-within .ws-ctab-x:hover {
color: var(--ws-ink);
opacity: 1;
}
.dark .ws-ctab:hover .ws-ctab-x:hover,
.dark .ws-ctab:focus-within .ws-ctab-x:hover {
background: rgba(255, 255, 255, 0.24);
}
/* Where the strip's scope ends and the pane's begins. */
.ws-ctabs-rule {
flex: 0 0 1px;
width: 1px;
height: 20px;
margin: 0 2px;
background: var(--ws-line);
}
@media (prefers-reduced-motion: reduce) {
.ws-ctabs {
scroll-behavior: auto;
}
.ws-ctab,
.ws-ctab-x {
transition: none;
}
}
/* ── Collapse tabs ───────────────────────────────────────────────────────
A collapsed pane leaves one narrow button with the same 30px surface pill
and reversed chevron permanently visible. The strip remains the generous
hit target; the pill makes that single action unmistakable. */
.ws-tab {
position: relative;
width: 30px;
background-color: var(--ws-rail);
background-image: var(--ws-grain);
box-shadow: inset -1px 0 0 0 var(--ws-line-soft);
color: var(--ws-ink-mute);
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease);
}
.ws-tab::after {
content: '';
position: absolute;
left: 50%;
top: 50%;
width: 7px;
height: 7px;
border-top: 1.75px solid currentColor;
border-right: 1.75px solid currentColor;
opacity: 0.72;
transform: translate(-65%, -50%) rotate(45deg);
transition:
color 160ms var(--ws-ease),
opacity 160ms var(--ws-ease);
}
.ws-tab::before {
content: '';
position: absolute;
left: 50%;
top: 50%;
width: 22px;
height: 30px;
border: 1px solid var(--ws-line);
border-radius: 9px;
transform: translate(-50%, -50%);
background: var(--ws-raised);
box-shadow: var(--ws-lift-sm);
transition:
border-color 160ms var(--ws-ease),
box-shadow 160ms var(--ws-ease),
transform 160ms var(--ws-ease);
}
.ws-tab.ws-pane-last::after {
transform: translate(-35%, -50%) rotate(-135deg);
}
.ws-tab:hover {
background-color: var(--ws-tint);
color: var(--ws-ink);
}
.ws-tab:hover::after,
.ws-tab:focus-visible::after {
color: var(--ws-ink-soft);
opacity: 1;
}
.ws-tab:focus-visible::before {
border-color: var(--ws-accent-thread);
box-shadow:
0 0 0 3px var(--ws-accent-wash),
var(--ws-lift-sm);
}
.ws-tab:active::after {
color: var(--ws-accent);
}
.ws-tab:active::before {
border-color: var(--ws-accent-thread);
transform: translate(-50%, -50%) scale(0.96);
}
.ws-tab:focus-visible {
outline: none;
box-shadow:
inset -1px 0 0 0 var(--ws-line-soft),
inset 0 0 0 1px var(--ws-accent-thread);
}
.ws-tab-label {
position: absolute;
width: 1px;
height: 1px;
overflow: hidden;
clip: rect(0 0 0 0);
white-space: nowrap;
clip-path: inset(50%);
}
/* Full-screen classroom playback hosts this existing control inside the
classroom chrome. It inherits the rail row's quiet ghost language: compact
by default, a soft tint only on hover. The ArrowLeft's shaft is hidden so
the visible mark is the requested chevron-left from the same Lucide path. */
.ws-root [data-testid='workbench-return'] {
height: 28px;
gap: 4px;
padding: 0 7px;
border-radius: 7px;
color: var(--ws-ink-mute);
background: transparent;
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease);
}
.ws-root [data-testid='workbench-return']:hover {
color: var(--ws-ink);
background: var(--ws-tint);
}
.ws-root [data-testid='workbench-return']:focus-visible {
outline: none;
color: var(--ws-ink);
background: var(--ws-tint);
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
.ws-root [data-testid='workbench-return'] svg {
width: 14px;
height: 14px;
}
.ws-root [data-testid='workbench-return'] svg path:last-child {
display: none;
}
/* ── Slim navigation rail ────────────────────────────────────────────────
The collapsed navigation: icons only, same glass, same hover pill. */
.ws-mini {
width: 60px;
}
.ws-mini-btn {
display: inline-flex;
align-items: center;
justify-content: center;
width: 34px;
height: 34px;
border-radius: 9px;
color: var(--ws-ink-mute);
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease);
}
.ws-mini-btn:hover {
background: var(--ws-tint);
color: var(--ws-ink);
}
.ws-mini-btn[aria-current='true'] {
background: var(--ws-accent-wash);
color: var(--ws-accent);
}
.ws-mini-btn:focus-visible {
outline: none;
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
/* ── The wordmark, as a link ─────────────────────────────────────────────
It has to read as pressable without becoming a button beside the PRO pill,
so the affordance is the rail's own hover tint and nothing else. */
.ws-home {
border-radius: 8px;
transition: background-color 160ms var(--ws-ease);
}
.ws-home:hover {
background: var(--ws-tint);
}
.ws-home:focus-visible {
outline: none;
box-shadow: 0 0 0 2px var(--ws-accent-thread);
}
/* ── Utility overflow ────────────────────────────────────────────────────
The ⋯ matches its neighbours in the foot, which are SiteHeader's 32px
round ghost buttons — a differently-shaped sibling would read as a
different kind of thing. */
.ws-util-btn {
display: inline-flex;
align-items: center;
justify-content: center;
width: 32px;
height: 32px;
border-radius: 9999px;
color: var(--ws-ink-mute);
transition:
background-color 160ms var(--ws-ease),
color 160ms var(--ws-ease);
}
.ws-util-btn:hover {
background: var(--ws-tint);
color: var(--ws-ink);
}
.ws-util-btn:focus-visible {
outline: none;
box-shadow: 0 0 0 1px var(--ws-accent-thread);
}
/* The panel is portalled to `.ws-root` and positioned against the viewport,
so it is never clipped by the rail's scrolling body nor re-based by the
rail's backdrop-filter. `z-index` clears the rail (10) and the panes. */
.ws-pop {
position: fixed;
z-index: 60;
display: flex;
align-items: center;
gap: 2px;
width: max-content;
padding: 4px 6px;
border-radius: 12px;
border: 1px solid var(--ws-line);
background-color: var(--ws-raised);
background-image: var(--ws-grain);
box-shadow: var(--ws-lift);
animation: ws-pop-in 140ms var(--ws-ease) both;
}
.ws-pop:focus {
outline: none;
}
/* One row, in the same order and at the same size as the foot cluster it
continues — the overflow is the rest of that cluster, not a second kind of
menu. */
.ws-pop-item {
display: inline-flex;
align-items: center;
gap: 2px;
}
@keyframes ws-pop-in {
from {
opacity: 0;
transform: translateY(4px);
}
to {
opacity: 1;
transform: none;
}
}
/* SiteHeader's own dropdowns open downward and are right-aligned, which at the
foot of a full-height rail is off the bottom of the window and — for a
control sitting 100px from the left edge — off the left of it too. Inside
the rail's utility clusters they open upward and hang to the right instead.
Radix-based menus position themselves and are unaffected. */
.ws-utils [class~='top-full'] {
top: auto;
bottom: 100%;
left: 0;
right: auto;
margin-top: 0;
margin-bottom: 8px;
}
/* ── Drag to reorder ─────────────────────────────────────────────────────
The drop affordance is one accent line where the row will land — not a ghost
card, not a gap that opens: the tree must not reflow under a pointer that
has not committed to anything yet. */
.ws-row-wrap.ws-drop-before::after,
.ws-row-wrap.ws-drop-after::after {
content: '';
position: absolute;
z-index: 3;
left: 2px;
right: 2px;
height: 2px;
border-radius: 2px;
background: var(--ws-accent);
box-shadow: 0 0 0 3px var(--ws-accent-wash);
pointer-events: none;
}
.ws-row-wrap.ws-drop-before::after {
top: -1px;
}
.ws-row-wrap.ws-drop-after::after {
bottom: -1px;
}
/* The carried row lifts above the tree without opening a layout gap. Transform
keeps the gesture compositor-only; the opaque raised surface lets the
shadow read against both the light and dark rails. */
.ws-row-dragging {
z-index: 4;
border-radius: 8px;
transform: scale(1.018);
box-shadow: var(--ws-lift-sm-hover);
}
.ws-row-dragging .ws-row {
background-color: var(--ws-raised);
color: var(--ws-ink);
}
/* A folder is not an insertion point, it is a container: it lights up whole
rather than growing a line. */
.ws-drop-into {
background-color: var(--ws-accent-wash);
box-shadow:
inset 0 0 0 1px var(--ws-accent-thread),
0 0 0 3px var(--ws-accent-wash);
}
/* The blank ground under the tree, as the "leave the folder" destination.
Invisible at rest — it reads as the list's own breathing room — and stated
only while a course is actually being carried. Finder's rule: you can always
drop something back out into the window's empty space, but the space does not
advertise itself when there is nothing in your hand.
STICKY, and inert until armed. On this account the tree is 139 rows deep, so
a zone that only existed at the end of the scrolled content would be
unreachable mid-drag (nothing auto-scrolls under a carried row) — the gesture
would exist for short lists only. Pinned to the panel's foot it is always
where the hand already is. It therefore overlaps the last rows, which is
exactly why it takes no pointer events at rest: a transparent band that ate
their clicks would be a bug wearing an affordance's clothes. Armed, it takes
the pointer AND an opaque ground, so it reads against whatever scrolled
under it. */
.ws-ground-drop {
position: sticky;
bottom: 0;
z-index: 2;
flex: 1 1 auto;
min-height: 44px;
margin: 6px 2px 0;
display: flex;
align-items: center;
justify-content: center;
border: 1px dashed transparent;
border-radius: 10px;
pointer-events: none;
transition:
background-color 160ms var(--ws-ease),
border-color 160ms var(--ws-ease);
}
.ws-ground-drop-armed {
pointer-events: auto;
border-color: var(--ws-line);
background-color: var(--ws-raised);
}
/* Two classes, so the destination's own accent wash outranks the armed ground
above rather than losing to it on source order. */
.ws-ground-drop.ws-drop-into {
background-color: var(--ws-accent-wash);
}
.ws-ground-drop-label {
font-size: 11px;
line-height: 1.3;
color: var(--ws-ink-mute);
}
/* Cursor and selection are locked on the document while a drag is in flight,
because the pointer travels far outside the row it started on. */
[data-ws-dragging] {
cursor: grabbing !important;
user-select: none !important;
}
@media (prefers-reduced-motion: reduce) {
.ws-pop {
animation: none;
}
.ws-home,
.ws-util-btn {
transition: none;
}
.ws-resize-thread,
.ws-resize-thread::after {
transition: none;
}
}