Executive dashboard
Operational overview, chart and data table, confidence display, sample reports and portfolio allocation.
No products match your search.
Program management & acquisition
Engineering
General productivity
Network infrastructure
Physical & logical environment model
End-user technology
Geospatial
Operations & logistics
Infrastructure control plane
Federation platform
Operational overview, chart and data table, confidence display, sample reports and portfolio allocation.
Reference navigation, reading, bookmarks and in-place page authoring with explicit drafts and publication.
Team pages, linked knowledge, page history, discussions and shared in-place authoring.
Channels, independent message histories, filtering, reactions, drafts and references to work.
Mailbox navigation, message reading, folders, search and non-modal rich-text composition.
Configurable saved boards, issue details, click and drag moves, column mappings and advisory WIP limits.
Source navigation, editable buffers, local saves, search, terminal visibility and isolated previews.
Repositories, branches, files, commits, pull requests, review gates and in-place editing.
Modal control room with Identity, Current scenario, Scenario library, Runtime and Evidence; separate non-modal Scenario Guide. Virtual-credential changes end the session without replacing requests. PostgreSQL stores the structured catalog, instances and run evidence. The embedded catalog is read-only preview when dependencies are unavailable.
13 acceptance scenarios and one diagnostic detector. Runtime gates, first-install seeding, stable scenario revisions, and negative credential conditions are explicit. Automated action descriptors use rendered controls and independent SQL/DOM checks. This build’s external QA is not part of scenario-coverage totals.
Sigma / build & design review
The implementation, its evidence, and the decisions still being worked through. All included in this HTML.
Establish supporting contracts through the product workflows, including the first approver for a commercial organization.
A prospective first OA supplies the proposed contract inside their initial account request. Submission creates a linked contract and appointment request transactionally. The ISO reviews the contract independently, then approves the account and scoped OA appointment. Pending, returned, declined, future-dated, and expired contracts do not authorize account approval. Contract approval alone creates no account or authority.
Vesta → Contracts offers drafts, submitted revisions, return and correction, ISO approval or decline, and current availability. An active OA may submit for organizations they represent; a prospective commercial OA has only the narrow permission to submit or correct their own linked proposal. ISO decisions include the actor, relevant authority, and submitted revision. Approved contract details remain read-only.
Commercial initial account sponsorship pins the independent contract approval. Its end date participates in the ISO's maximum-date calculation and is checked again at provisioning and activation. Public user identifiers remain unassigned until provisioning. Account approval, provisioning, and usable OA authority remain distinct. The government initial-operator path is preserved.
Abbreviation is optional. Organization relationships still use a stable identifier. Empty abbreviations neither conflict with each other nor produce an empty line or separator in the directory.
Eight new acceptance scenarios cover the commercial full lifecycle, focused commercial first OA, ordinary OA contract submission, the complete government onboarding foundation, contract return/resubmission, decline, future-dated approval, and failed submission/retry. Twelve new flow records or revisions reuse task-level credential, contract-detail, review, provisioning, and organizational setup behavior. Earlier definitions and runs retain their original revision pins.
Schema 7 adds independent contract submission and decision records, contract revisions, and identifiers. Opening this release does not reset working data, the loaded scenario, or Evidence. Existing approved reference contracts remain historical records rather than being assigned invented review history. A pending commercial initial-OA request without contract information must be returned for correction; it cannot silently use the government exemption. Previously active accounts are not retroactively revoked by this upgrade.
The current build QA view identifies actual checks and test environments. The browser workflow harness uses the production charts and SQL repository over a test-only SQLite adapter with direct projection. This does not establish PostgreSQL locking, PGlite durability, or real Datastar delivery. The delivered application retains the real runtime and contains no test-only database or projection adapter. Live scenario Evidence is not prepopulated with build-test results.
Approved-contract amendments, renewals, early termination, support withdrawal, routine OA nomination and authority-management workflows remain subsequent increments. Contract registration here is an approved basis for access, not procurement or contract award.
Replace the next seeded prerequisites with product workflows: registered organizational scope and ISO-sponsored initial OA authority.
Establish an organization and its first OA runs 43 checkpoints. It begins at the existing out-of-band provisioned empty environment, executes the first-ISO commissioning flows, registers the organization through the ISO's Vesta screen, and passes the actual ISO and organization outputs to the candidate's request. The completed world has one organization, one ISO, one active scoped OA, and no contracts or ordinary applicant requests.
The ISO records account sponsorship and OA appointment as distinct decisions. The account-support date and authority date are explicit, bounded by existing policy and the sponsor's account/authority. A system-authorized provisioning command assigns public identifiers; the candidate then accepts responsibility and activates the account and scoped grant. No authority is conferred by the test-person catalog, the approver hint, submission, or approval alone.
Vesta → Organizations registers a Government or Commercial organization with its allowed affiliations and shows initial approval coverage. Prospective OAs use Request initial OA appointment and the four-field ISO identity hint. Vesta → Appointments provides a persistent queue, separate sponsorship/appointment confirmations, date helper, return with explanation, decline, and next-request actions. The normal contract-backed account workflow remains intact.
The production path remains Datastar, internal channel/SSE, interpreted commands, and transactional PGlite mapping. New SQL tables hold initial-operator requests, decisions, and provisioning records; scoped grants reuse the authority table. Working data, loaded scenarios, catalog revisions, historical run plans, credentials, and earlier evidence are not reset when this build opens. The original Origo style blocks are retained.
Five focused cases cover the ordinary appointment, return and resubmission, explicit decisions and dates, provisioning rollback/retry, and a wrong ISO hint. Their upstream commissioned state is prepared by command replay, explicitly not credited as earlier UX execution. The full composed journey executes those prerequisites through actual controls. Failed runs keep the guide at the failed checkpoint.
Build QA distinguishes pure domain checks, production SQLRepository calls over a test-only SQLite driver with direct UI projection, and browser layout/interaction checks. Those do not establish actual PGlite/Datastar execution, production authentication, directory integration, or server security. No test-only database or projection adapter is included in this HTML. See QA evidence for recorded checks and limitations.
The newly appointed OA can subsequently submit the first contract for ISO review. Contract authoring, ordinary OA nomination, authority revocation/recovery, policy editing, and operational metrics are outside this revision.
2026-09-07.2 · Implementation delivered; real-runtime acceptance is pending.
Existing Vesta controls → genuine Datastar backend actions → scoped MessageChannel/SSE → interpreted application and request transitions → PGlite transactions → committed display projection. The Workroom compiler, transition and actor core and its loader/transport patterns were adapted from the retrieved source. Vesta's chart definitions, relational mapping and storage controls are new.
Saved mode opens an IndexedDB-backed PGlite database under a single-tab ownership lock. Temporary mode is an explicit choice, not an automatic fallback. Working and scenario records have separate SQL data-space keys. Scenario replacement does not replace saved working requests, requirements, feedback or other applications' data. Actor switching is a separate lab operation that continues a request.
The old Vesta JSON snapshot is imported only into empty working data, with validated identities, request IDs, revisions, history and command receipts. The source is retained. An exact SQL read-back mismatch aborts the import. Export/import and offline-runtime packaging live under Account → Preferences → Access data. Inspect SQL records reads the tables rather than the UI projection cache.
This build environment could not retrieve the pinned PGlite/Datastar assets. No claim is made that the full database-backed browser journey or offline runtime has passed here. Pure chart/domain tests, an explicitly labeled recording adapter for SQL parameter mapping, and Chromium failure-path checks are separate evidence categories. Check live runtime executes real SQL and a Datastar projection probe after the dependencies load in your browser. M2.1 is not accepted merely because the shell opens.
The ordinary account journey and the existing Origo/Sigma surfaces are preserved. Commit and dropped-projection faults are observable. No CUE/Rego/Lean or Rust compatibility proof, live IdP, server-side authorization, or external directory provisioning is implemented. Events are an audit trail, not full event sourcing. The manual/human-speed/machine-speed scenario runner remains M2.3.
Public entry, applicant request, high-volume reviews, and a shared Account / Identity surface.
Public sign-in simulation; three-step request drafting; request status; Organizational Approver and ISO queues with 16 requests; explicit per-request contract support; return/decline notes; provisioning and activation; identity identifiers and expirations; simulated credential enrollment and removal; local cryptographic passphrase generation.
The Sigma header, workspace switching, appearance controls, profile-based ranking, attention, feedback and original Origo sources remain. Existing Preferences controls were moved into Account without creating a second implementation.
Entry-state loading and actor switching are confined to this optional Review Center. Commands, repository evidence and DOM revisions are observable. Manual/auto-human/auto-machine scenario execution remains the next integration, not a completed feature.
No real account, authentication, hardware key, password verifier, or external provisioning service is created. Provider interactions and credential bindings are labeled local simulations. The passphrase generator performs actual local random generation; phrases are not stored or attached to an account. Website source could not be fetched again during this revision; new interaction rules are identified under Vesta Access.
Object-row presentation, explicit profile bindings, source-time semantics, and the settled workspace-shell pattern in Origo.
Home, full collections, universal attention, and Origo specimens now use the same three row variants. The object, action reason, location, and content preview have separate roles. Working text and essential metadata retain their sizes; complete-row fitting determines capacity.
My work and attention use the configured profile's explicit source bindings. Conversation read markers have a profile-specific namespace. Display-name changes do not change authorship; unknown bindings do not silently inherit KS or the seeded mailbox. Existing domain stores remain shared local fixtures, not protected multi-user data.
Activity is a latest-source projection. Only UTC/offset timestamps are treated as dated events. Relative display labels are not converted into invented dates. New chat messages and repository commits record their real creation time; historical sample records remain undated.
The sidebar, identity, header, search, feedback and universal attention controls are retained. The guide and gallery use live shared controls and source-linked rows. Acceptance contracts are independent of Review Center; the QA evidence records external browser checks, not Scenario Lab runs or formal proofs.
Removing Review Center removes these reports and the review installer, not the shell contracts, source adapters, search, feedback, attention or design-system content.
Outline attention utility, global source count, non-modal panel, stable updates, contextual navigation, and independent removal of Review Center. Current eligibility rules and source records are unchanged.
Header/canvas alignment, compact intrinsic command-strip sizing, Home-only deferred-update placement, and runtime platform shortcut hints. Panel rows, source stores, sidebar measurement and profile accounting remain unchanged.
Removed only the account button’s visible Preferences subtitle. Existing action, accessible name and responsive sizing are retained.
Profile-scoped launch counters, a single entry-count boundary, settled sidebar ranking and a simplified graphical account control. No changes to the original Origo CSS, source-view content or sidebar-fit measurement algorithm. Complete older records below are historical.
Complete M1 records plus explicit changes for this revision. The older ledger is archived below rather than rewritten as a new historical record.
| Area | Treatment | Revision |
|---|---|---|
| Preview delivery | Adapted | The complete application is losslessly gzip-packed inside this HTML. Expansion requires no network, module import, persistent storage or child iframe. A bundled decoder fallback is included; expanded byte length and completion marker are checked. |
| Initialization and routing | Hardened | Initialization waits for DOM readiness and is guarded against duplicate entry. Opaque and about: preview documents use the existing session-memory history adapter without attempting to rewrite their embedding URL. |
| Architecture v9 reader | Repaired | The inaccessible nested iframe is replaced with a style-isolated Shadow DOM reader. The original v9 source string and both SVG figures are retained exactly. |
| Review Center | Integrated | Release overview, all QA results and limitations, the full porting ledger, formatted notes, and session diagnostics are directly readable in one view. Full original M1 evidence is archived and not presented as fresh revision evidence. |
| Editor networking | Explicit in previews | The original ProseMirror/Loro setup remains available. An embedded preview does not start downloading editor dependencies until the user explicitly enables setup from Review Center → Runtime. Top-level operation remains on-demand; no substitute editor is introduced. |
| Product/capability navigation | Discussion only | Record the user’s direction: an Origo authoring/studies surface, a Sigma activity front door, and reusable Sibyl capabilities hosted in product contexts. The M1 one-owner routing remains temporary. This revision does not implement new operational workflows or claim that Workflow belongs only to Artifex. |
Full design system and application source · 994,510 bytes
SHA-256 ed528e16fd6d6c0c596897f3bbdd7d206cf069051a7a1f7a21c6219e5d162f1d
Authoritative product names and boundaries · 39,176 bytes
SHA-256 e865a28b145e913859f24426b5f59e92aa822213294b69c9d53a2edd3be0a0e8
| Area | Treatment | What changed — and what did not |
|---|---|---|
| Origo visual system | Preserved | All 13 source stylesheet blocks are byte-for-byte unchanged. The MIT notice in the embedded AmbientCSS core is retained. Sigma styles are added in a separate, scoped block. |
| Design guide | Preserved | The full original guide, its section navigation, surface studies, implementation contracts, handoff checklist and dynamically inserted authoring guidance remain in the document. |
| Gallery and studies | Preserved | Existing primitive demonstrations, structural checks, theme/lighting controls, curated palettes, scene presets, import/export, entity pickers and collaborative workbench are retained. |
| Eight application concepts | Preserved | Original markup and domain behavior remain. Sidebar visibility, shell labels and product-aware routes are integration changes; sample repository names and illustrative data are not globally renamed. |
| Product registry | Added | All 11 product names, groupings, domains and responsibilities come from v9. Card summaries paraphrase v9. Temporary monograms and prototype placement are M1 decisions, not approved product branding or new whitepaper claims. |
| Home and product landing pages | Added | A launcher, capability search, product switcher, recent-object list and honest architecture-only states. No generated operational metrics or simulated service health. |
| Navigation adapter | Adapted | Product-prefixed routes coexist with legacy Origo hashes. Browser/session-history fallback, object links, scroll restoration and existing authoring-exit behavior are retained. |
| Appearance compatibility | Adapted | The existing Origo appearance schema is retained. Original ten-surface presets receive defaults for new Sigma surfaces; exported presets include those new surfaces. |
| Whitepaper v9 | Embedded | The entire supplied HTML is stored intact in the document, including both SVG figures and source references. The reader now uses a style-isolated Shadow DOM surface, without a nested browsing context. Heading targets and external-link handling do not alter the embedded original. |
| Offline authoring export | Adapted | The existing offline builder exports the combined Sigma document under a descriptive Sigma filename after the real editor runtime loads. It does not bundle browser-local user content. |
| Storage | Preserved / added | Origo data remains under its original storage keys. Sigma keeps only launcher history in its own namespace. Storage availability is detected; this is not a backup, migration or multi-user synchronization system. |
| Scenario Lab / cost model | Not integrated | M2 and M3 respectively. No substituted runner, new numerical model or fake passing scenario is presented in M1. |
Inspect this document. These checks are not Scenario Lab, a usability audit or formal verification.
Not run in this session. Checks cover registry, retained views, unchanged styles, source embedding and navigation bindings.
Preserve the source; establish Sigma Home and v9 product boundaries.
Port the existing runner and connect real UX to candidate requirements.
Port the substantial PGlite cost model and its time-based planning experience.
Author, review and version acceptance criteria and execution evidence.
Exercise one coherent handoff through normal product actions.
Establish supporting contracts through the product workflows, including the first approver for a commercial organization.
A prospective first OA supplies the proposed contract inside their initial account request. Submission creates a linked contract and appointment request transactionally. The ISO reviews the contract independently, then approves the account and scoped OA appointment. Pending, returned, declined, future-dated, and expired contracts do not authorize account approval. Contract approval alone creates no account or authority.
Vesta → Contracts offers drafts, submitted revisions, return and correction, ISO approval or decline, and current availability. An active OA may submit for organizations they represent; a prospective commercial OA has only the narrow permission to submit or correct their own linked proposal. ISO decisions include the actor, relevant authority, and submitted revision. Approved contract details remain read-only.
Commercial initial account sponsorship pins the independent contract approval. Its end date participates in the ISO's maximum-date calculation and is checked again at provisioning and activation. Public user identifiers remain unassigned until provisioning. Account approval, provisioning, and usable OA authority remain distinct. The government initial-operator path is preserved.
Abbreviation is optional. Organization relationships still use a stable identifier. Empty abbreviations neither conflict with each other nor produce an empty line or separator in the directory.
Eight new acceptance scenarios cover the commercial full lifecycle, focused commercial first OA, ordinary OA contract submission, the complete government onboarding foundation, contract return/resubmission, decline, future-dated approval, and failed submission/retry. Twelve new flow records or revisions reuse task-level credential, contract-detail, review, provisioning, and organizational setup behavior. Earlier definitions and runs retain their original revision pins.
Schema 7 adds independent contract submission and decision records, contract revisions, and identifiers. Opening this release does not reset working data, the loaded scenario, or Evidence. Existing approved reference contracts remain historical records rather than being assigned invented review history. A pending commercial initial-OA request without contract information must be returned for correction; it cannot silently use the government exemption. Previously active accounts are not retroactively revoked by this upgrade.
The current build QA view identifies actual checks and test environments. The browser workflow harness uses the production charts and SQL repository over a test-only SQLite adapter with direct projection. This does not establish PostgreSQL locking, PGlite durability, or real Datastar delivery. The delivered application retains the real runtime and contains no test-only database or projection adapter. Live scenario Evidence is not prepopulated with build-test results.
Approved-contract amendments, renewals, early termination, support withdrawal, routine OA nomination and authority-management workflows remain subsequent increments. Contract registration here is an approved basis for access, not procurement or contract award.
Replace the next seeded prerequisites with product workflows: registered organizational scope and ISO-sponsored initial OA authority.
Establish an organization and its first OA runs 43 checkpoints. It begins at the existing out-of-band provisioned empty environment, executes the first-ISO commissioning flows, registers the organization through the ISO's Vesta screen, and passes the actual ISO and organization outputs to the candidate's request. The completed world has one organization, one ISO, one active scoped OA, and no contracts or ordinary applicant requests.
The ISO records account sponsorship and OA appointment as distinct decisions. The account-support date and authority date are explicit, bounded by existing policy and the sponsor's account/authority. A system-authorized provisioning command assigns public identifiers; the candidate then accepts responsibility and activates the account and scoped grant. No authority is conferred by the test-person catalog, the approver hint, submission, or approval alone.
Vesta → Organizations registers a Government or Commercial organization with its allowed affiliations and shows initial approval coverage. Prospective OAs use Request initial OA appointment and the four-field ISO identity hint. Vesta → Appointments provides a persistent queue, separate sponsorship/appointment confirmations, date helper, return with explanation, decline, and next-request actions. The normal contract-backed account workflow remains intact.
The production path remains Datastar, internal channel/SSE, interpreted commands, and transactional PGlite mapping. New SQL tables hold initial-operator requests, decisions, and provisioning records; scoped grants reuse the authority table. Working data, loaded scenarios, catalog revisions, historical run plans, credentials, and earlier evidence are not reset when this build opens. The original Origo style blocks are retained.
Five focused cases cover the ordinary appointment, return and resubmission, explicit decisions and dates, provisioning rollback/retry, and a wrong ISO hint. Their upstream commissioned state is prepared by command replay, explicitly not credited as earlier UX execution. The full composed journey executes those prerequisites through actual controls. Failed runs keep the guide at the failed checkpoint.
Build QA distinguishes pure domain checks, production SQLRepository calls over a test-only SQLite driver with direct UI projection, and browser layout/interaction checks. Those do not establish actual PGlite/Datastar execution, production authentication, directory integration, or server security. No test-only database or projection adapter is included in this HTML. See QA evidence for recorded checks and limitations.
The newly appointed OA can subsequently submit the first contract for ISO review. Contract authoring, ordinary OA nomination, authority revocation/recovery, policy editing, and operational metrics are outside this revision.
The Identity selector now reads credentials from the current Vesta identity rather than substituting catalog defaults. Earlier first-signup references are retained; no requests, decisions or account records are reset or rebound.
Identity and the virtual credential check use the same validation. Invalid inputs remain blocked, with a direct Choose another credential action on the bridge. Reload the selected credential once to replace an older virtual selection; a scenario reset is not required.
Build QA below distinguishes production code exercised with controlled fixtures from the still-unverified PGlite/Datastar integration. Earlier run evidence is retained under its original build.
2026-09-07.2 · Implementation delivered; real-runtime acceptance is pending.
Existing Vesta controls → genuine Datastar backend actions → scoped MessageChannel/SSE → interpreted application and request transitions → PGlite transactions → committed display projection. The Workroom compiler, transition and actor core and its loader/transport patterns were adapted from the retrieved source. Vesta's chart definitions, relational mapping and storage controls are new.
Saved mode opens an IndexedDB-backed PGlite database under a single-tab ownership lock. Temporary mode is an explicit choice, not an automatic fallback. Working and scenario records have separate SQL data-space keys. Scenario replacement does not replace saved working requests, requirements, feedback or other applications' data. Actor switching is a separate lab operation that continues a request.
The old Vesta JSON snapshot is imported only into empty working data, with validated identities, request IDs, revisions, history and command receipts. The source is retained. An exact SQL read-back mismatch aborts the import. Export/import and offline-runtime packaging live under Account → Preferences → Access data. Inspect SQL records reads the tables rather than the UI projection cache.
This build environment could not retrieve the pinned PGlite/Datastar assets. No claim is made that the full database-backed browser journey or offline runtime has passed here. Pure chart/domain tests, an explicitly labeled recording adapter for SQL parameter mapping, and Chromium failure-path checks are separate evidence categories. Check live runtime executes real SQL and a Datastar projection probe after the dependencies load in your browser. M2.1 is not accepted merely because the shell opens.
The ordinary account journey and the existing Origo/Sigma surfaces are preserved. Commit and dropped-projection faults are observable. No CUE/Rego/Lean or Rust compatibility proof, live IdP, server-side authorization, or external directory provisioning is implemented. Events are an audit trail, not full event sourcing. The manual/human-speed/machine-speed scenario runner remains M2.3.
Public entry, applicant request, high-volume reviews, and a shared Account / Identity surface.
Public sign-in simulation; three-step request drafting; request status; Organizational Approver and ISO queues with 16 requests; explicit per-request contract support; return/decline notes; provisioning and activation; identity identifiers and expirations; simulated credential enrollment and removal; local cryptographic passphrase generation.
The Sigma header, workspace switching, appearance controls, profile-based ranking, attention, feedback and original Origo sources remain. Existing Preferences controls were moved into Account without creating a second implementation.
Entry-state loading and actor switching are confined to this optional Review Center. Commands, repository evidence and DOM revisions are observable. Manual/auto-human/auto-machine scenario execution remains the next integration, not a completed feature.
No real account, authentication, hardware key, password verifier, or external provisioning service is created. Provider interactions and credential bindings are labeled local simulations. The passphrase generator performs actual local random generation; phrases are not stored or attached to an account. Website source could not be fetched again during this revision; new interaction rules are identified under Vesta Access.
Attention rows separate the object title, outstanding reason and context. Continue working includes the recorded last-opened time and draft flag. Activity uses event titles instead of bold message bodies, with separate excerpts and actor/context metadata. All row icons use an 18px outline with consistent stroke. Final visible rows no longer double the footer separator.
Profile/source bindings are declared in the local configuration. An explicit authorProfileId or assigneeProfileId takes precedence over legacy bindings. The mailbox binding describes the existing local mailbox, and team-attention opt-in describes the existing local engineering queue. None of this grants authorization. Existing unscoped conversation read markers are not attributed to a user; the new profile-specific markers start fresh.
Activity rows with complete ISO timestamps sort ahead of undated snapshots. Time-only and relative strings are retained as source labels in tooltips, not interpreted as dates. No sample timestamp is fabricated. Activity remains a source snapshot rather than a complete history.
The new Workspace shell guide and gallery family exercise the same global controls and row renderer used in working applications. The twelve previously curated design areas and all thirteen original stylesheet blocks are preserved.
Scenario Lab is not yet integrated. Shared criteria are candidate acceptance contracts, with external browser evidence separately recorded. Native ChatGPT preview, native downloaded-file persistence and optional ProseMirror/Loro dependency success require their own environment checks. The new source bindings do not migrate all legacy domain data into a protected multi-user architecture.
The outline bell and its count are a permanent header utility immediately left of feedback. The total is drawn from the same source projection as Sigma Home and is not scoped to the active workspace or filter. The compact panel exposes All, Action needed and Unread with source links and the existing full queue.
Inspecting the panel never acknowledges or completes work. Source state changes update the badge; hovered or keyboard-focused rows wait for an explicit refresh. Closing returns focus, following work uses the normal route path, and mobile content behind the full-width panel is temporarily removed from keyboard navigation. Search and feedback do not stack over attention, and feedback drafts survive switching.
No remote notification feed or operating-system notifications are connected. This revision retains the existing local attention eligibility rules, including the seeded user, sample pipeline subscriptions and pending reviews. It does not introduce production authorization, recipient routing, or notification-event deduplication.
Current tests are scoped to this revision. Original Origo styles, sidebar sizing, launch accounting, source stores, search semantics and dashboard panels are retained. The small feedback hover treatment now matches the new outline utility.
Search titles, saved body content, messages, files, work items, and Origo guidance. Private drafts are excluded. New matches are resolved from current records rather than a copied dashboard list. Product scope is explicit; a shared capability retains its host context.
Search routes preserve the query and scope. Search-only transitions do not manufacture product launches or overwrite the last working object. Existing body renderers, datasets, appearance machinery and sidebar fitting are retained.
Feedback drafts and local records are profile-keyed. Failed persistence is disclosed as session-only. A captured context never follows subsequent navigation silently. Saved history and export remain usable after Review Center is removed.
The source dataset has no full authorization model. Search association fixtures are not permission enforcement. Successful rich-editor loading and the native ChatGPT app remain separate environment checks.
Priority 4, framing only. Panel information hierarchy is deliberately deferred.
Search now aligns with the bounded Home canvas, including centered ultrawide layouts. The empty leading desktop header cluster and the static “Your workspace” label no longer consume space. The section tabs remain on narrow and shallow Home layouts.
The global header has a 56px minimum with a fine pointer, or 60px with a coarse pointer. Search retains its 40px/44px minimum target and existing font size. Intrinsic header height is measured and shared with content surfaces when text needs more space.
The existing updates control appears in the header only when Home has a deferred update. It does not shift the panel rows under the pointer. Search advertises Ctrl K or ⌘ K according to the runtime platform; modified browser/debug shortcuts and text composition are not intercepted.
Current checks are scoped to these changes and relevant regressions, not a full rerun of all earlier releases. The native app preview and optional external editor runtime require separate verification.
Label-only refinement.
The account button now displays only the avatar and profile name. Preferences remains the click/keyboard action and is retained in the tooltip and accessible name. The accepted layout and launch accounting are unchanged.
Priority 3 only. The dashboard, header, width, workspace identity and measured-fit algorithm remain unchanged.
One entry into a different application through normal navigation counts once, after successful route restoration. A sidebar shortcut has one delegated input path. Selector, directory, search, resumption and attention links use the same accounting boundary. Origo is included. Local views, reloads, history restoration, invalid routes and Preferences/Review detours are not additional launches.
Counter shards are keyed to the configured stable profile ID and open document instance. Tabs do not overwrite each other’s counters. Display names are not keys. Unscoped legacy counts are retained but not assigned to any profile: corrected ranking starts fresh. Equal counts sort alphabetically. The visible order settles on entry into Sigma; rendering, resizing and other-tab storage events cannot reshuffle it while choosing an application.
A neutral inline profile icon, full name and the short Preferences label replace initials plus a clipped subtitle. Click or keyboard activation opens Preferences directly. The compact rail retains the icon and an accessible name. No extra menu or remote image request is used.
The current HTML has one configured local profile. This release does not introduce sign-in, account switching, authorization or cross-device synchronization. Two-profile verification uses isolated configuration fixtures against the same storage. Only launch personalization is newly profile-scoped; domain data, authoring identity and other preferences retain their existing single-user model. When local storage is unavailable, launches remain in memory and Preferences reports the limitation.
Priority 2 only: distinguish switching from identity.
The selector uses a stable Sigma mark and Workspaces label in every application. The name/version block is non-interactive, without border, fill, shadow, hover or an extra keyboard stop. The Home route label is shortened and redundant workspace-name captions are suppressed. A compact monogram prevents narrow-rail truncation; full names and versions remain in the accessible description.
The sidebar-fit controller, ranking, profile storage, account, header and product content are retained. All previous QA records remain historical evidence. Optional Review Center export remains available.
Priority 1 only: complete-row fitting; reserved directory/account controls; keyboard-safe resize and explicit extreme-height fallback. Current evidence is under QA. Older notes below are historical.
Build 0.1.2. This is a navigation and integration release; M2 Scenario Lab and M3 cost planning remain deferred.
Sigma Home now projects attention and activity from the existing application state. Product switching resumes working locations; product overviews are separate destinations. Origo contains the full guide, gallery, Theme Lab, Lighting Lab and concept library. Basic appearance selection remains in preferences.
Wiki, email, chat and workflow can render in a host product without redirecting the user into Sibyl. Artifex, Opifex and Sibyl use the same workflow board and inspector, with explicit domain profiles. A personal work queue shows references to assigned work, not copies.
Drag-and-drop now observes the same illustrative completion rule as the status picker. Dismissed entity pickers cannot leave delayed outside-click listeners that close a newer picker. Navigation preserves working input focus and selection instead of replacing them with the global toolbar focus. QA evidence records the exercised paths and distinguishes blocked or unverified environments.
The thirteen original Origo style blocks, original guide, gallery, repository, communication, authoring and application content are retained. The original theme and lighting control nodes are moved into the Origo surfaces, not recreated. The v9 source and all prior review evidence remain embedded.
The original twelve engineering work items retain their identifiers. Two personal and two operational sample items were added to exercise context-specific workflow; their provenance is recorded in the ledger. No real IT operations or messaging service is connected.
See QA evidence for the exact current-run checks and environment. In-app candidate requirements are not approved requirements or formal proofs. Read receipts and home activity are local projections. Product context is not a security boundary.
The rich authoring dependencies remain pinned and on-demand. Browser-local data is not bundled into HTML source exports. Native ChatGPT app preview execution requires a user-side check. Cross-product links do not implement authorization or federation.
Live diagnostics, separate from the recorded build tests. Copy these details when reviewing a preview-specific problem.
The shell and all reading surfaces start without external requests. The original ProseMirror/Loro authoring implementation can load its pinned dependencies on demand. In an embedded preview, explicitly enable that setup here; the host may still request or deny network access.
No replacement editor is used when setup is unavailable. Enabling setup does not send your document content to a model or external service; it requests the editor libraries.
Loading runtime details…
No new network dependency on startup. The launcher, design system, gallery, concepts and whitepaper are embedded. Rich-text authoring retains the pinned ProseMirror/Loro setup. In embedded previews, explicitly enable downloads under Runtime before using it. The inherited offline builder requires those libraries to load successfully; it exports the full application but not the local content database.
Datastar 1.0.3 → scoped MessageChannel / SSE → Workroom’s exclusive statechart interpreter → PGlite 0.5.8. The existing Vesta screens remain the working interface. No scenario runner is claimed here.
Working and scenario records use separate SQL data-space keys. Loading a fixture replaces only the scenario data space. Participant switching retains the same request; it is a development control, not authentication. New browser sessions reopen working data.
Import previous records before starting a new working request. Import never overwrites a nonempty working data space.
Not run in this browser.
Open the runtime, then inspect the database. There is no substitute SQL adapter.
Effects execute on the browser thread behind the channel boundary; there is no worker, Rust service or server-side authorization here. The event table is an audit trail, not a full event-sourcing proof. A single database owner is enforced for saved storage.
Prepare a local starting point, then use the real product controls. These are fixtures and inspection tools, not an automated scenario runner.
Switching the fixture actor is a privileged lab action, not authentication. It preserves the requests and handoff history. Return here with Ctrl / Command + Shift + L.
These controls affect only Vesta Access. A dropped projection is an intentional defect; a successful command is not evidence that the screen is correct.
The lifecycle captured in the prior read-only Vesta Lab review is retained: identity evidence, draft, organizational review, ISO review, approval, and activation after provisioning. ISO contract-support disposition remains explicit. The hosted lab did not respond to fresh reads on September 7, 2026; its Rust statechart, guards, and templates were not available for a source-code port.
Queue interaction, the synthetic Northstar fixture population, return/decline rules, credential confirmation simulation, and the two-step provisioning representation are UX implementation choices in this document. They are not attributed to uninspected source semantics. Invitation binding, accountability handoff, and privileged-access nomination are not implemented in this pass.
Vesta runs a versioned local command/repository/projection adapter. It now uses PGlite and the ported Workroom interpreter behind real Datastar SSE. It does not run the hosted lab’s Rust interpreter, CUE/Rego guards, or a real identity provider. Data in a client-only file is not protected by production authorization. All provider and credential bindings are visibly simulated; the passphrase generator uses real browser cryptographic randomness with a curated 1,050-word vocabulary.
The Scenario Testing POC separates fixture setup, pure domain decisions, repository effects, rendered projections, and independent assertions. Preserve that separation when adding manual, human-speed, and machine-speed execution. All execution modes must use the same DOM actions and checkpoints. No run is claimed by loading a fixture.
Commands have stable IDs, expected request revisions, actor IDs, and reset epochs. Domain rejection leaves records unchanged. A transaction-callback failure requests rollback. A transport or durability interruption may have an uncertain outcome; reload SQL state before retrying. Fixture reset increments the epoch. Inspect request and repository revision attributes in the DOM independently of VestaAccess.repository(). Generated passphrases are excluded from stored state, command logs, and snapshots.
The SQL reset boundary is the Vesta scenario data space only; saved working requests are separate. Review evidence, requirement curation, appearance, feedback, and the other product data must survive. One workflow can switch actor repeatedly without resetting its request.
Load First contact. Continue with CAC / PIV and confirm the demonstration identity. On the request, use Northstar Program Office and morgan.reyes@example.org as routing. Add a work purpose and future requested end date, then submit. Use Ctrl / Command + Shift + L to open the same workflow as Morgan, then Rowan, then Casey. Do not reload a fixture between actors. Return as Jordan to inspect activation and Identity.
Sample data · Focus the chart, then use ← / → to inspect points.
| Date | Actual | Baseline | Difference |
|---|
| Program | Owner | Health | Progress | Next gate |
|---|---|---|---|---|
Nexus modernizationPlatform transformation | R. Chen | On track | Sep 12 | |
AI enablementEnterprise capability | M. Patel | On track | Sep 18 | |
Data convergenceInformation architecture | A. Brooks | Watch | Sep 09 | |
Client experienceService redesign | S. Ortiz | On track | Sep 24 |
A layered platform architecture that keeps product experiences independent while sharing identity, policy, event, and data capabilities underneath.
Each workspace surface can feel purpose-built without fragmenting the underlying product. Shared services provide continuity; local components preserve task focus.
The experience is organized around a common application shell and a set of bounded workspaces. Dashboard, knowledge, communication, workflow, and development surfaces all consume the same identity, authorization, search, and event backbone.
The result is a cohesive product family rather than a collection of embedded tools. Users retain consistent navigation and lighting cues while the information density and interaction model adapt to the work at hand.
Workspace boundaries should be visible but not disruptive. Navigation state, global actions, notifications, and lighting remain stable. Toolbars, content layout, and local interactions change according to the work surface.
interface WorkspaceSurface {
id: "dashboard" | "docs" | "chat" | "mail" | "workflow" | "code";
shell: "shared";
density: "calm" | "standard" | "dense";
primaryAction: Action;
capabilities: Capability[];
}
Shared tokens define contrast, spacing, typography, state, and semantic elevation. AmbientCSS then renders those semantic decisions through a coherent material and lighting model rather than becoming the source of the UX semantics itself.
Kyle,
Leadership aligned on the proposed MVP path during this morning’s readiness review. The portfolio remains healthy overall, and the schedule and cost posture support proceeding with the next decision gate.
Three decisions were confirmed: retain the shared application shell, treat the event backbone as a platform capability, and sequence the first workload migration behind the identity boundary validation.
I’ve attached the leadership readout and the action register. Please reply with any corrections before 1500 so we can publish the final record.
— Maya
Kyle,
Leadership aligned on the proposed MVP path during this morning’s readiness review. The portfolio remains healthy overall, and the schedule and cost posture support proceeding with the next decision gate.
Three decisions were confirmed: retain the shared application shell, treat the event backbone as a platform capability, and sequence the first workload migration behind the identity boundary validation.
I’ve attached the leadership readout and the action register. Please reply with any corrections before 1500 so we can publish the final record.
— Maya
The platform boundary update now reflects the notification service decision, security review comments, and the operational ownership model.
The proposed design keeps message delivery behind a stable contract while allowing individual workspaces to choose local notification behavior.
The final diff and the verification notes are linked in the architecture workspace.
Confirming our understanding of the support period and the two funding categories discussed yesterday.
Personnel support continues through the transition horizon, while enterprise business system support covers access and hosting commitments. The fixed funding amount is unchanged by an earlier release of support personnel.
Our team is available for a 30-minute reconciliation session tomorrow morning.
The latest pass uses physical depth to clarify hierarchy rather than decorate every surface.
Dashboard metrics retain raised cards and recessed tracks. Documentation uses a matte reading surface, chat emphasizes conversational layers, and the code editor becomes the densest workspace.
I recommend keeping the current balance and validating contrast in the lightest and darkest scene presets.
The quarterly authorization evidence package is complete. Automated evidence passed validation and the two manually attested controls have been accepted.
Owner certification is required by Friday at 1700. No open findings block certification.
Tuesday, 15 September
09:00–11:00 UTC
Hover, focus, or press. The light belongs to the control.
A raised edge invites. A pressed edge confirms.The same controls used in the workspace, with complete behavior and difficult content.
Default, interaction and unavailable states
control.buttonA reviewer must be assigned first.
Save to exercise the pending state.
Persistent hover, pressed and focus appearances above are labeled specimens. Tab through the actual buttons to test real keyboard focus.
Label, help, error and confirmation
control.fieldReadable and selectable, but not editable.
Make the next step explicit
feedback.contentKeep the region’s shape stable while content is pending.
! The preview is unavailable. Your draft is still here.
Choose “No records” in the content selector to test the table’s empty state and disabled bulk-selection control.
Labels accompany every signal
feedback.statusZero is a value; “not recorded” is missing data. Do not use the same visual treatment or wording for both.
Controls that initiate, confirm, or change work.
Data entry and choice controls with restrained physical cues.
Progress, system state, and compact status communication.
The small physical vocabulary used to build larger products.
Use a tonal boundary and one semantic accent. Do not add elevation unless the callout can move or receive input.
Two useful patterns for everyday business workflows.
Actual controls and source-linked rows used by the applications.
Origo creates coherence across distinct business applications by giving every surface the same visual physics, semantic tokens, and hierarchy rules. It favors clarity and restraint, then uses light, depth, color, shape, and motion to make interaction easier to understand.
Origo uses a quiet reading field and a selective, physical interaction layer. Roughly 80% of the surface supports reading; the actionable 20% catches the light. These are design proportions, not numerical quotas. A strong material treatment must predict an action, a state change, or a temporary layer.
A raised button presses down. A focused input gains a lit rim. A selected tab sits above its track. A draggable card lifts during movement. Menus float over the work they affect.
Do not add hover glows to static KPIs, glass to document pages, or a floating shadow to every card. Preserve non-color cues and a visible focus outline. Never require hover to discover a primary action.
| Role / state | Required behavior | Material recipe |
|---|---|---|
| Reading content | Stable during light changes; no motion on hover. | Opaque, planar, clear text. |
| Action at rest | Recognizable before interaction. | Shallow body, directional edge and compact cast shadow. |
| Hover or keyboard focus | More inviting without moving neighboring content. | At most 1px lift; stronger edge and reflection. Focus also has an outline. |
| Pressed | Immediate, reversible physical feedback. | Body lowers; shadow tightens. No bounce. |
| Selected | Persistent state, not permanent hover. | Accent-tinted material, explicit marker and ARIA state. |
| Disabled or read-only | No action invitation. | No lift or moving reflection; preserve a readable label. |
| Floating controls | Above the work; dismissible and keyboard reachable. | Frosted background, directional rim, clear occlusion. Do not nest glass layers. |
Glass is reserved for action surfaces and floating controls. It is a browser approximation built from translucent fills, blur, directional reflections, and AmbientCSS shadows—not optical refraction. Reduced-transparency and forced-colors fallbacks use opaque surfaces. Reduced motion removes lift and animated reflection.
Start with a complete, touch-ready single-pane experience. Add persistent navigation and supporting panes only when the primary content remains usable. A phone is not a squeezed desktop, and an ultrawide monitor is not permission to stretch paragraphs.
| Available space | Layout responsibility | Interaction responsibility |
|---|---|---|
| Narrow / touch | One working pane. Page, channel, inbox, and file navigation use dismissible drawers. Wrap action groups together. Keep the composer visible. | Primary actions are always discoverable without hover. Use 44px action targets where practical; preserve text labels or explicit accessible names. |
| Tablet / compact desktop | Add a compact app rail, then a local navigation pane. Do not squeeze an editor or reader to fit optional context. | Maintain keyboard order and return focus after closing navigation. A resize must not change the chosen item or lose a draft. |
| Wide / ultrawide | Bound paragraphs to roughly 70–80 characters. Align message history with its composer. Tables, boards, and code can use a wider working frame. | Keep actions adjacent to the object they affect. Context remains optional. Never enlarge every gap merely because the screen is wider. |
| Short / landscape | Reduce toolbar overhead, not the reading font. Make the terminal collapsible. Every work scroller must retain useful height. | Dialogs scroll internally. On-screen keyboards must not cover the active composer; pinch zoom must remain available. |
Use container width for dashboard and gallery columns. Test long labels, empty states, and mixed content at the thresholds, not only at device presets.
Page-level overflow checks are only a first check. Inspect the actual header, composer, action row, drawer, and dialog. Preserve intentional internal scrolling for boards, tables, and source code.
These principles are ordered. When principles compete, favor the earlier one.
Use coherent light, a visible edge, and slight elevation to invite interaction. Pressing lowers the control; focus is just as visible as hover.
About 80% of the working field should support reading without demanding attention. The interactive 20% earns stronger light and material response; this is a balance, not a quota.
Passive content is planar. Actions sit above it. Inputs are shallow wells. Temporary controls float. Do not make read-only content look like a button.
Highlights, cast shadows, and pressed states share the same direction. Pointer response stays local to the active control; it never tilts the entire application.
The material changes around the label, not through it. Keep text, focus, boundaries and state readable in every supported theme and lighting preset.
Share action behavior across applications. Let documentation stay calm, conversations stay readable, workflow cards feel movable, and repository tables stay precise.
Separate accent fill, accent text and selection surfaces. Derive them from a deliberate harmony, then test their roles in real rendered components.
Respond quickly, move only a small distance, and settle. Give touch and keyboard users the same affordance. Respect reduced motion and never depend on hover alone.
Origo treats the interface as one illuminated scene. Designers specify semantic roles and geometry; the rendering system resolves the final material response.
Depth is intentionally narrow. Most interfaces need a canvas, a planar surface, a shallow raised surface, a recessed region, and a temporary overlay.
| Level | Use | Visual treatment |
|---|---|---|
| Inset | Inputs, tracks, wells, contained data regions | Concave edge, low tonal contrast, no floating shadow |
| 0 · Planar | Canvas, reading surfaces, tables, persistent panes | Tonal separation and rules; no elevation shadow |
| 1 · Raised | Interactive cards, buttons, selected tools | Shallow edge and restrained shadow |
| 2 · Floating | Pinned controls, floating toolbars, high-priority objects | Clear separation; use sparingly |
| 3 · Overlay | Menus, dialogs, command palettes | Strongest depth; always temporary |
The default material. It supports long sessions, dense information, and broad surface areas without competing with content.
Default for reading and structureA modest highlight that clarifies shape and interaction. Suitable for primary controls, indicators, and selected objects.
Use for focused interactionGloss, glow, or stronger saturation belongs only on a key state, active control, or compact data indicator.
Only with a functional reasonA named hue harmony sets relationships. Contrast-constrained role colors make those relationships usable. Accent fill, accent text and selected backgrounds have different luminance requirements and must not share one color token.
Use role tokens such as accent, surface, text, success, warning, and critical. Never bind a component to a literal hex value.
Keep neutral surfaces close in hue and separate them primarily by lightness. This preserves calm while maintaining hierarchy.
Reserve the primary accent for focus, selection, active progress, and the highest-priority action in a local region.
Status meaning must survive grayscale. Pair color with text, iconography, position, or shape.
Validate every generated theme in light and dark mode. The relationship is the design; neither mode is a secondary skin.
--accent-fill--on-accent--accent-text--accent-surface--panel--ink--muted--control-borderUse --accent-fill with --on-accent for filled controls; --accent-text for links; and --accent-surface for selected rows. Keep link underlines in reading text.
A pleasing swatch is not a contrast test. Do not reuse a pale button fill as small link text on a light background. Never use color as the only status cue.
Use a 12–32px scale at the default 16px root. Working text is 14–15px; sustained reading is 16px. Compact mode changes row spacing and working text by one step, never document text or metadata below 12px. Use approximately 1.65 line height for messages and 1.8 for articles.
| Token | Size | Use |
|---|---|---|
--text-meta | 12px | Metadata only |
--text-label | 13px | Supporting labels |
--text-body | 14px | Controls, tables and code |
--text-content | 15px · 14 compact | Chat and email |
--text-reading | 16px | Documents and Wiki |
--text-title | 24px | Page headings |
--text-display | 32px | Home headings / focal metrics |
Motion connects user action to system response. It should be brief, interruptible, and removed when the operating system requests reduced motion.
Hover, press, selection, focus, and small disclosure. Prefer opacity, transform, and tonal changes.
Panels, menus, view transitions, and objects moving between containers. Preserve a visible origin and destination.
Animate only when it helps the user understand direction or magnitude. Do not replay on every refresh.
Shared lighting may change deliberately in a lab or theme context, but routine work should not feel visually unstable.
Accessible states are designed with the component, not added after visual approval.
Target at least 4.5:1 for normal text and 3:1 for large text, icons, controls, and essential boundaries.
Every interactive element has a visible focus ring that is not dependent on elevation or subtle color change.
Provide at least a 44 × 44 px touch target even when the visible icon is smaller. Dense pointer layouts still need 32 px minimum hit areas.
Never communicate status by color, glow, or depth alone. Include a label, icon, pattern, or explicit value.
Use the same role tokens and component classes in applications and the Gallery. Create an exception only when the task requires it, and document why it cannot use an existing contract.
| Token | Value | Use |
|---|---|---|
--space-1 / 2 / 3 | 4 / 8 / 12px | Icon–label relationships, related controls and small groups. |
--space-4 / 5 / 6 | 16 / 20 / 24px | Rows, control groups and object padding. |
--space-8 / 10 | 32 / 40px | Article sections and major regions. |
--radius-control | 8px | Buttons, inputs and search fields. |
--radius-object | 12px | Cards, linked work and grouped content. |
--radius-overlay | 16px | Dialogs and laboratories. |
| Intent | Component | Required behavior |
|---|---|---|
| Perform an action | .primary-btn / .soft-btn | Native button, visible focus, pending state when needed, a truthful result. |
| Navigate to content | Named link or navigation item | A stable location; Back restores the previous page and reading position. |
| Collect a value | .form-field | Persistent label, help/error relationship, validation without erasing input. |
| Manipulate a work item | .issue-card | Pointer and keyboard entry; editable status as an alternative to dragging. |
| Interrupt for a decision | Modal dialog | Named dialog, initial focus, contained tab order, Escape, return focus. |
| Tune appearance while observing | Nonmodal laboratory | Move focus in; allow tabbing out; Escape returns to its launcher. Do not block the work. |
Wrap message and issue text. Truncate navigation labels only when the full label remains available. Scroll a wide data table within its region. At 820px and below, let local navigation become a dismissible drawer.
Do not shrink body text to preserve a composition. Do not let a long filename widen the page. Do not hide the only way to change status when dragging is unavailable.
A handoff includes default, hover, pressed, keyboard focus, disabled and—when relevant—loading, success, error, empty and read-only states. Disabled actions explain their prerequisite nearby. A failed action preserves the user’s input and provides a specific recovery action.
Keep reading backgrounds stable. Let shared light explain a button’s edge, an input’s recess, an issue’s movability or a temporary overlay. Test each effect under Quiet and Sculpted with the same content.
Do not bevel every nested container. Do not let a theme experiment make body text faint. Do not raise a static paragraph simply because elevation is available.
Check all ten views in both modes; one generated high-chroma palette; one extreme light; keyboard-only completion; reduced motion; long and missing values; and a 320px viewport. Check focus and control boundaries manually. Theme Lab’s computed-color samples and Gallery’s structural checks are useful evidence, not complete accessibility certification.
Application identity and local navigation belong in the sidebar. The header provides workspace-scoped search, portfolio-wide attention, and feedback. Each product owns its working objects, local actions and domain rules; changing the host does not change an object's identity.
The shell surrounding this guide is the same shell used by the applications. These controls open its actual search, attention panel, feedback composer and workspace selector. They are not a second set of demonstrations.
Attention presents an object and an outstanding reason; resumption presents an actual working location; activity distinguishes what happened from its content preview. Use the shared row renderer and source adapters. Keep the object's route stable and never mark it read or complete merely because it appears in a summary.
Workspace identity: stable ID, name, publication version local navigation: capabilities and object routes search: published records with explicit workspace associations attention: source IDs, outstanding reasons, destinations actor bindings: stable profile mapped to source identities context: selection, draft, focus and return location
Search follows the workspace. Personal attention follows the person. Source bindings describe which local records correspond to that person; they do not authenticate them or grant access. A production source must enforce its own authorization before returning data.
Reserve the selector and account area before fitting complete navigation rows. On Home, show bounded previews with a clear route to the full collection. Short or narrow windows change composition before typography; extreme constraints use an explicit scrolling fallback. A temporary panel must not move the object under a pointer or discard keyboard focus.
Empty, unavailable and stale are different states. New records can update the attention count while an inspected list waits for refresh. Feedback keeps its original application context and distinguishes a saved local record from delivered feedback. Activity snapshots are not a full audit log: only complete timestamps support chronological ordering.
Handoff describes intent as well as appearance. A developer should understand the semantic role, states, and responsive behavior without reverse-engineering the mockup.
Choose a treatment per work surface in Lighting lab. It changes object thickness, elevation, and recess depth—not navigation, meaning, or text. Compare the same content, then export the complete scene as a portable JSON preset.
Minimal body height, near-planar containers. Start here for documentation, correspondence, and prolonged reading.
Subtle edges on manipulable objects, shallow wells, and a calm field. Start here for Wiki and conversation.
Stronger separation between cards, deeper tracks, and polished accents. Treat as an exploration, not a universal default.
Use polished material for a small indicator or control, not paragraphs. Keep broad reading surfaces matte. Frosted overlays need a stable backing so text remains legible. Reduce or remove depth wherever it competes with the content.
This single-file exploration bundles the AmbientCSS core used here: calibrated lighting, edges, surfaces, recesses, and matte/shiny/glass treatments. Grain textures and the React control package are not included.
Design system
Explore application patterns in their working contexts.
Origo / appearance
Author the shared appearance here. Basic appearance selection remains in Sigma preferences.
Shared tokens and actual Origo component classes.
Keep the content legible and the work in context. Light belongs to the controls, not to the paragraph.
Interactive specimen only · no product record is changed.
Changes use the existing shared appearance engine. Return to the exact product context you brought into Origo.
This study is not a live duplicate of your product screen.
Origo / light & material
Author the shared appearance here. Basic appearance selection remains in Sigma preferences.
Shared tokens and actual Origo component classes.
Keep the content legible and the work in context. Light belongs to the controls, not to the paragraph.
Interactive specimen only · no product record is changed.
Changes use the existing shared appearance engine. Return to the exact product context you brought into Origo.
This study is not a live duplicate of your product screen.
Use a fixed mode or follow this device.
A curated appearance applies a matched palette and lighting. The selected color mode remains in effect.
Application order uses your launch history.
Settings are stored in this browser when storage is allowed. The HTML is not a backup of browser-local documents, messages or drafts.
Enable rich-text editing for documents and messages. Setup may request permission to load the editor.