Conversation
Replace the developer-oriented architecture/basics/guides pages with a new docs structure written for non-technical platform users, starting with a comprehensive Communication section covering every channel (Console chat, Unify Meet, phone, SMS, WhatsApp, email, Slack, Teams, Discord, meetings), channel setup and costs, voice selection, assistant communication behavior, and recordings/transcripts.
…ction Document the workspace connection from the user's perspective: what it unlocks, the Console connect flow (providers, feature grants, OAuth), inbox access, calendar, Drive/OneDrive/SharePoint files with access rules, contacts and tasks, and security/revocation. Cross-link from the Communication email page and the landing page.
…nnel logos Replace the legacy green-cube PNG with the current isometric Unify mark plus a Space Grotesk wordmark (glyphs outlined to paths, light/dark variants) generated from the branding repo. Move the theme from the old #009600 green to the brand teal/purple palette, and restyle card icons with varied brand token colors and real multicolor logos for Slack, Microsoft Teams, and Google Meet.
Document integrations from the user's perspective: what connecting an app unlocks, Native vs Third-party apps, how integrations differ from the workspace connection and communication channels (including the dual-role Slack case), the connect flows (OAuth, API key, custom secrets, account labels, multiple accounts), per-tool permissions with presets and approvals, credential handling, and everyday use including scheduled tasks and apps outside the gallery.
Document tasks from the user's perspective: the kinds of tasks (scheduled, triggered, on-demand, background), chat-driven creation with natural-language schedules and event triggers including Test it, run-time behavior (automatic wake-up, silent-by-default execution, reporting on connected channels, timezone and credit rules), and the read-only Tasks tab with statuses and run history. Cross-link from the behavior, workspace, and integrations pages.
Document how the assistant learns from natural interaction: the post-work review that saves playbooks (Guidance) and reusable skills (Functions), watching live work in the Actions pane including the retrieval and storage steps, browsing both libraries in the Brain section, and how to teach effectively — corrections mid-work, "remember this", screen-share walkthroughs, documents, and the onboarding teach-by-correcting demo. Cross-link from the tasks and behavior pages.
Document the Dashboards area (adopting the upcoming Canvas name) from the user's perspective: the two building blocks — structured data tables the assistant builds from APIs, files, integrations, and its own computations, and live dashboard tiles wired to those tables — plus the Console Data and Dashboards panes, sharing via tile/dashboard links, and worked recipes for API monitoring, KPI boards, combined sources, persisted intermediate work, and derived variables.
Document the assistant's managed Ubuntu/Windows desktop from the user's perspective: what the assistant does on its machine (browser sessions, native apps, persistent files including chat attachments), watching the live desktop during a Unify Meet call and taking mouse-and-keyboard control for joint work, and the machine itself — OS choice at hire, persistent per-assistant storage across sessions, voice-first calls with the desktop unlocking mid-call, and privacy/isolation guarantees. Cross-link from the Unify Meet and Integrations pages.
Document linking a user's own machine from the user's perspective: what the Unify desktop app enables on macOS, Windows, and Ubuntu, the install-and-link flow including the optional macOS login password, filesystem access with the never-overwrite-originals guarantee and the new shell-command capability with its guardrails, and the behavior and revocation model — explicit-request-only control, confirmation before consequential actions, per-level revocation, and the outbound secure tunnel. Cross-link from the Their Computer pages.
Document teams from the user's perspective: grouping people and assistants within an organization, the two membership models — each person's hidden T-W1N automatically mirroring its owner's team access, versus publicly visible assistants joining teams directly — and the headline use case of shared context: personal-by-default memory with team pools for knowledge, guidance, functions, tasks, data, dashboards, and credentials, scoped retrieval across personal plus member teams, and the boundaries that keep sensitive material unit-specific. Includes team administration and the managed org-wide sharing team. Re-point existing shared-workspace mentions at the new section.
Document the two teammate identity models from the user's perspective: the automatic private T-W1N that acts as its owner through their own account and platform-routed contact pools, versus hired teammates with dedicated Google/Microsoft accounts (own gated file and app access, own Teams/Slack profiles), their own contact details and integration credentials, and the ability to liaise across human and virtual stakeholders under per-contact response policies. Covers the Onboard Teammate flow, first-day setup, End contract, delegation between T-W1N and colleagues, and the recommended twins-first adoption path.
Deepen the Console chat page with the pane's real capabilities (voice notes transcribed into the composer, 32MB attachments and camera capture, reactions, TTS playback, inline call transcripts and embeds, conversation search with filters, spending-block states). Add a Contacts page covering the directory, per-contact detail (channels, bio, rolling summary, should-respond and response policy), and personal-plus-team merging. Rework the transcripts page around the Transcripts pane: channel rail, Threads vs Feed layouts, cross-channel search, and how the three windows onto the record relate.
Describe the gallery's three sections (Connected apps, Needs attention, Available apps), the card badges and connection statuses, the per-assistant scoping of the tab, and the recently shipped bring-your-own-OAuth section for admins — custom branding and scopes, apps that require a custom OAuth app, and the redirect-URI setup flow.
Expand the monitoring page to reflect the Console Tasks tab precisely: the definition-plus-run-history card layout with its structured fields, the Run Detail view, the in-product help popover, footer counts, and a new section on what the tab is for in practice — post-creation checks, did-it-run glances, failure triage, and workload audits.
…ards panes Explain what each dedicated pane is for and how to use it: the Data pane as the read-only audit window into ingested tables (directory tree, sorting, pagination, team groupings, verification workflows) and the Dashboards pane as the viewer for built views (picker with dashboards and standalone tiles, summary card, per-tile live badges and refresh, pop-out links, downloads, and auto-refresh while the assistant works).
Add a dedicated Knowledge page explaining the facts library — what lives there, how it fills through documents and conversation, the pane itself, upkeep, and the Knowledge/Guidance/Functions division. Deepen the Actions page with the pane's controls (time window, search, expand all) and concrete uses, and describe the Guidance and Functions panes' actual affordances: split reader with scope/tag filters, the Learned/Primitives filter, and the function detail drawer.
Note the auto-opening dialog for first-time users, which fields are required, the role label's placement in the assistant list, and the workspace step's explicit choose-or-skip gate before submission.
Add a How it works page to the Your Computer section explaining the app's background services and outbound-only tunnel, and disambiguating the two mechanisms: Meet's Share your screen is a view-only browser feed (which does show the user's machine), while the linked desktop is a control connection that is never displayed in the Console. Documents the assistant's resolution order (active share first, linked desktop when no share) and how to supervise work on a linked machine.
Deep technical reference for the open-source unify integrations subsystem, aimed at engineers extending or debugging the code: the three integration lanes (provider-backed, native packages, SDK plus secrets), the Builtins catalog and bootstrap seeding, provider tool materialization into FunctionManager with per-app hashing and the sync lifecycle coordinator, the governed execution path and approval loop, native package discovery and secret-gated enablement, the external-app SDK pattern with the OAuth capability proxy, extension guides, and a test map. Includes three generated architecture diagrams and GitHub links to every referenced module.
Developer-facing documentation of the open-source communication code: the gateway/runtime split with the envelope wire contract and local-vs-hosted transport seams, the channels and adapters packages with per-module references, the ConversationManager event model, mediums, slow-brain scheduling and CommsPrimitives send pipeline, and the dual-brain voice architecture including fillers, barge-in, speaker enrollment, and hang-up semantics. Includes three generated architecture diagrams matching the repo README's visual style, with GitHub source links throughout.
Document the open-source task system for developers building on the unify runtime: the Task model and its invariants (schedule XOR trigger, task_id/instance_id split, the typed RRULE subset), the lifecycle state machine with execute-time re-arm cloning, the TaskScheduler manager surface and validation funnel, the four Tasks storage contexts with team destination routing, the three activation lanes (scheduled live, two-stage trigger matching, headless offline), steerable ActiveTask execution, symbolic entrypoints with the offline certification gate, and local vs hosted scheduling. Includes three generated architecture diagrams matching the unify README's visual grammar, with GitHub file references throughout.
Developer-facing documentation of the open-source workspace code: the end-to-end OAuth lifecycle from gateway callbacks through Orchestra secrets to the runtime, the split token-storage model and sandbox isolation in runtime_oauth and SecretManager, the provider proxy's request classification, file-access policy engine and response filtering, the WorkspaceEmailManager send identities, and the Gmail/Outlook/Drive/SharePoint gateway channels with watches and the scope model. Includes three generated architecture diagrams matching the repo README's visual style, with GitHub source links throughout.
Document the open-source team-scoping machinery for engineers: ContextRegistry destination resolution with the shared-table and membership gates, SHARED_SCOPED_TABLES and authorship stamping, the write-one/read-all invariant with the transcript fanout exception, federated search merge semantics, per-manager scope differences (secrets vaults, task execution ownership, dashboard data_scope), the SESSION_DETAILS membership flow with live AssistantUpdateEvent refresh, the accessible-teams prompt block, coordinator lifecycle tools, and extension recipes — with three generated architecture diagrams in the house style.
Five developer-facing pages covering the open-source learning machinery with GitHub source links and generated architecture diagrams: the two-phase act() and discovery-first tool policy, FunctionManager (primitives registry, compositional functions, venvs, execution routing), GuidanceManager (CRUD, previews, pinning, builtins), StorageCheck (skill librarian loop, store_skills, task entrypoint certification), and Knowledge/Memory (NL tool loops over dynamic tables, 50-event consolidation window, extension points). Also carries the concurrently authored Tasks developers page referenced by the shared navigation.
…internals Code-level documentation of the open-source data and visualization layers: the shared base/real/simulated manager pattern with docstring-driven primitive discovery, the table-as-context storage model with ContextRegistry destinations and federated multi-root reads, the query engine (filter expressions, embedding-backed search, reduce, ephemeral and materialized joins), the DAG-based ingest pipeline, and the DashboardManager tile system — binding types, on_data contract, dry-run validation, token lifecycle, the render-time bridge, and destination vs data_scope. Includes generated architecture diagrams and GitHub source links throughout.
…rnals Code-level tour of the open-source runtime's user-desktop stack for developers: the UserDesktopLink session state and per-user resolution for shared assistants, the primitives.computer.user_desktop factory and handle with per-call consent assertions, the on-demand SFTP file channel with its staging mirror, timestamped .unity-edits writebacks and tiered excludes, the user_desktop execution surface for shell and Python, the three-layer consent model with live revocation events, the prompt layer that encodes the behavioral rules, instructive tests, and extension points. Includes three generated architecture diagrams matching the README's visual style, with all references linked to files on the public unify repo.
Document the open-source unify computer stack for developers: the primitives.computer namespaces with readiness gating and secret injection, the ComputerBackend/ComputerSession/MagnitudeBackend HTTP layer, the agent-service protocol and session modes, the VM desktop composition and desktop_url contract, the AssistantDesktopReady / FileSyncComplete lifecycle converging on apply_managed_desktop_ready, rclone bisync workspace sync, execution surfaces with Windows PowerShell routing, and the screen-share/remote-control event plumbing including fast paths and the screenshot pipeline. Includes three brand-style architecture diagrams and GitHub source references throughout. Also wires the pending Learning Developers nav entries.
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Workflows to automatically generate PRs for you. |
magic-marty
approved these changes
Jul 6, 2026
magic-marty
left a comment
magic-marty
left a comment
There was a problem hiding this comment.
LGTM — promoting staging docs rewrite to production.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
stagingtomainfor production deployment at https://docs.unify.aiTest plan
mainmint.jsonon staging