DottaandPaperclip
d38cdf261a
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Tasks and Inbox are the main entry points for that work. > - Earlier changes made their data reads and row counts smaller. > - Those changes remain, but shared controls now pull in the whole task-detail bundle. > - A cold open can also read a remembered organization before selecting the one in the URL. > - This pull request separates collection code and selects the addressed organization first. > - It warms task code after collection paint, so the next task open stays fast. ## Linked Issues or Issue Description Refs #17756, #18323, #16849, #17041, #17983, #17960, #18461. **What happened?** Tasks and Inbox still use their optimized index reads and row budgets. However, small shared controls and hooks were placed in the task-detail chunk. A collection had to download that whole chunk. The HTML also preloaded it on every route. The original production profile loaded about 6.5 MB of decoded JavaScript before useful rows, or 1.9 MB transferred with compression. Company bootstrap could select a different remembered organization before Layout selected the URL's organization. This started unnecessary company reads. A retained multi-company browser fixture reproduced two attention-count requests for different companies. **Expected behavior** Show correctly scoped, stable collection rows without waiting for task-detail code. Keep compact reads, bounded rendering, direct task preloading, immediate New Task input, and fast follow-on navigation. **Steps to reproduce** 1. Use a production UI build with Coordinator enabled and two active organizations. 2. Remember one organization, then cold-open Tasks or Mine in the other. 3. Inspect startup script and company requests before the first useful row. The baseline UI revision is `91fdf02e95`. This is a UI change. It has no API, schema, migration, or authorization changes. ## What Changed - Group the static collection dependency trees above the task-detail group. Keep the existing startup and task grouping strategy. - Preload task detail and its static dependency closure early on direct task and agent-chat URLs. Keep collection aliases out of that preload. - Fetch the immutable task chunk at low priority after a collection paints, or on a same-origin task-link hint. After a successful immutable response is fully read, module-preload the cached chunk so it can compile without executing. A failed HTTP probe skips module preload. - Prefer an addressed active organization during bootstrap. Preserve manual selection, archived-company handling, and account-scoped company lists. - Preserve a task reading anchor when a programmatic scroll event reports an unchanged offset during restore reflow. Strengthen the Back journey to wait for a completed collection navigation. - Add production-browser checks for delayed indexes, correct company scope, stable row geometry, failed speculation, and early direct-task preload. Update collection verification notes and the feature map. ## Verification Final source: `e402ab7fc6`; baseline UI: `91fdf02e95`. - All **234 focused unit cases** and **44 production-browser journeys** passed on the final source: 33 task journeys and 11 collection journeys. Coverage includes compact history, saved cards, draft persistence, avatar continuity, failed-probe recovery, delayed data, pagination, deep links, desktop/mobile and reading-position restoration through Back. Desktop and mobile screenshots were inspected. - Workspace typecheck, full production build, map checks, token gates and diff checks passed. The default local `pnpm test:run` still stops before executing tests because the existing chat-test line selector reports `No test found` (1,816 skipped). This is not a passed full local test suite. Focused tests and remote CI remain separate evidence. - Greptile reviewed the exact final commit at **5/5**, with zero open review threads. Final code CI passed. The complete check inventory is **55 successful, 3 skipped, 1 neutral**. The neutral result is the separate Paperclip Review: its agent finished without completing an assessment, so it is not a passing review score. Five alternating baseline/candidate pairs per scenario, Chromium 151.0.7922.34, 1440×900, 4× CPU, 50ms latency, approximately 9Mbps down / 1.5Mbps up. Cold cases use fresh browser contexts with normal cache writes enabled; warm cases use the same context and normal application query cache. Both previews use the server's immutable hashed-asset cache policy and uncached HTML, with real local feature settings and no request interception. All task-owned builds and suites were stopped during timing; other host activity was uncontrolled. Readiness requires useful rows and editable search, or the readable task thread and editable composer, followed by two animation frames. | Journey | Baseline median (range) | PR median (range) | Median paired saving | | --- | --- | --- | --- | | Cold Tasks | 3239 ms (3189–3333) | 2545 ms (2441–2682) | +752 ms | | Cold Inbox | 3207 ms (3203–3268) | 2679 ms (2502–3062) | +541 ms | | Warm Inbox | 717 ms (701–917) | 760 ms (745–768) | -47 ms | | Warm Tasks | 1150 ms (1126–1207) | 1136 ms (1119–1174) | +18 ms | | First task click | 943 ms (932–1008) | 1007 ms (976–1098) | -74 ms | | Back to Tasks | 584 ms (568–617) | 566 ms (559–575) | +10 ms | | Cold direct task control | 3389 ms (3377–3402) | 3415 ms (3395–3896) | -13 ms | All five pairs improved for each cold collection. The signed saving is baseline minus candidate for each pair; it is distinct from the difference of medians. Warm navigation is broadly similar. First task click was 74ms slower by median paired difference. Direct-task median paired difference was 13ms slower, with one slower 3.90s candidate sample retained in the range; this is not a claimed direct-task speedup. Completed script transfers by collection readiness fell from about **1.91MB to 1.14MB** (Tasks), and **1.90MB to 1.14MB** (Inbox). Task code warms after rows paint. At the test bandwidth, the removed 0.77MB critical transfer represents roughly 0.65s of download time; CPU and network overlap, so their savings are not added. Unthrottled loopback API control medians stayed about 17–18ms in both variants. All 30 cold samples had zero browser errors. Rejected experiments remain separate: automatic splitting produced over 200 startup requests; disabling recursive grouping slowed direct tasks; HTTP-only warming improved collection startup but slowed first task opens. The final correction preloads direct dependencies and compiles successfully cached code after paint. Routing disables HTTP caching, so intercepted runs were kept as diagnostic evidence rather than used for warm-cache acceptance; see [Playwright's cache behavior](https://playwright.dev/docs/api/class-page#page-route). Vite preview's default asset `no-cache` policy was also corrected to match the shipped server before final measurement. ## Risks - Bundle placement affects shared consumers. Browser checks cover collections and task navigation; the full PR checks also cover other routes. - Code warming still transfers the task chunk if the user stays on a populated collection. It moves that download past first paint and does not execute it. This is a startup latency improvement, not a claim that all eventual bytes disappear. - A direct click before warming finishes still loads code on demand. Direct task and chat URLs retain early module preload. - Local browser evidence does not establish a deployed staging speedup. This PR does not merge or deploy itself. ## Model Used OpenAI GPT-6 (Codex), with reasoning, tool use and local code execution. The context-window size is not exposed in this environment. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run the relevant tests locally and they pass (234 unit cases and 44 browser journeys; the full-command limitation is documented above) - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All code CI gates are green (the separate Paperclip Review is neutral because its agent did not complete an assessment) - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing>
14 KiB
navigation
| navigation | ||||||
|---|---|---|---|---|---|---|
|
Inbox, decisions, and search
People find work that needs their attention, distinguish unread activity from required decisions, archive handled items, and search across the company.
Feature ID: inbox-search.
Implementation: inbox, decisions, search.
Agent access: companion.
Sub-features
views:inbox-search/views— use Mine, Recent, Unread, Blocked, and broader task views.triage:inbox-search/triage— read or archive attention items without pretending the underlying task is complete.decisions:inbox-search/decisions— open the source of a decision and act in the right context.search:inbox-search/search— query company work and follow a result to its original record.toasts:inbox-search/toasts— receive live task notifications only for personal participation or explicit mentions.
How to get to it (user POV)
inbox
Use Inbox links or the corresponding task views when the combined list is enabled.
decisions
Open /decisions, then a queue at /decisions/queues/:key and its source item.
With Settings → Experimental → Coordinator workspace enabled, /decisions
is Needs You. Pending cards use the existing attention feed and canonical
resolvers. Tasks → Needs attention projects the same feed as distinct tasks.
Search, filter by agent/project, sort, and group by task/agent/project apply to
both open decisions and the visible Recent Decisions receipts. Successful
resolutions resize into compact receipts while the same contained Gather Light
effect used for task creation runs concurrently. The receipt stays in its card
position until that treatment finishes, then settles under Recent Decisions.
Motion follows the saved history result; failed saves retain the actionable card,
reduced motion settles immediately, and revisiting history does not replay it. The initial
24-hour display window is an experimental assumption, not data deletion.
Resolving an artifact decision does not itself complete its task. If the task
remains in review without a maintained reviewer or continuation, its separate
canonical stalled-review item stays in Needs You and Tasks → Needs attention.
Coordinator presents the existing recovery actions with Mark task done as
the task-level verdict, distinct from the saved artifact receipt.
Open Decision history on this surface for the paginated history sidebar; it merges answered questions and confirmations with Decisions-v1 outcomes.
search
Open /search or the navigation search entry.
toasts
Keep a company page open while another person or agent updates a task. See toast semantics for eligibility and suppression rules.
Driving it with Playwright and the Paperclip CLI
Preconditions: follow the baseline. Seed assigned, unread, blocked, and resolved tasks plus a pending decision. Record the combined Inbox/Tasks experimental setting.
inbox
Automated: archive routes cover server mutations. Task-list helpers cover search URLs, pagination, and presentation constants; they do not render the combined Inbox/Tasks page. Reading, archiving, and navigating that page remain manual checks.
Automated: Issue service regressions cover exclusion of historical automatic read cleanup from user participation, unread counts, and activity timestamps while retaining explicit user decisions and audit history.
Automated: Source regressions cover channel attribution, imported comments, direct participation, addressed requests, unread results, and access checks. Routine regressions cover fresh, coalesced, and skipped manual runs. The source regressions also create an addressed request through the service for a human assignee. They check unread lists, flags, and counts, plus exclusion of historical questions and closed requests while preserving authorization requests.
Manual: Seed GitHub, Slack, scheduled routine, and manual routine tasks. Attribution, imported comments, and reading alone keep them outside Mine. Direct board/API comments, human assignment, and open addressed requests admit them. Resolve the request and check that another valid reason is needed. Inbox discovery and Blocked hide unrelated automation by default; the Show agent-created tasks checkbox reveals it without subscribing the user. Reload and verify membership, unread badges, archive/resurfacing, and preference persistence.
Manual: Read an unread task, return to Unread, archive a handled item, and reload. Verify personal triage state changes while task status stays correct. Switch users and companies to check isolation.
Seed a completed task whose only activity under the viewing user is automatic read cleanup. Refresh Mine and Unread: the task must not appear through that participation, while the cleanup remains in activity history. Record an explicit user recovery decision on a separate task and verify it still counts. This browser journey remains manual.
Cold-open verification: load Mine with the task-index response delayed while
approvals/runs resolve. Only the stable list placeholder should appear until the
initial sources settle; a failed-run row must not jump when tasks arrive. Release
visible-row details separately and compare row/title bounds. Mine reads its own
100-row index, with parent/run context by ID; sidebar badges use the count endpoint.
Coordinator desktop and mobile navigation share /attention/summary for the
outstanding count. Needs You owns its complete feed; Tasks → Needs attention
owns a compact complete task-ID index. Navigation reuses either page response. Coordinator keeps the optimized Inbox even when
the instance Streamlined UI setting is off. The loopback browser checks in
collection startup exercise these
reads and the initial task row budget. They do not establish deployed latency.
Collection startup selects the addressed active organization before starting
company reads. Shared controls load separately from task detail. A painted
collection or task-link intent starts a low-priority code fetch; direct task URLs keep
early module preload. The collection
browser cases cover a remembered different organization and failed speculation.
The Inbox component checks cover the source
barrier and selected-source query; badge tests
cover unread counts. Browser timing and layout proof remain separate from these tests.
decisions
Automated: decision queues covers route behavior. End-to-end queue-to-source navigation is manual.
Manual: Open a pending decision, follow its source, resolve it through the supported control, and reload the queue. Confirm it retires only after the decision persisted and unrelated pending items remain.
Coordinator regression sources: canonical history exercises real PostgreSQL cross-source pagination, answered-question retention, company/private-task isolation, and private v1 targets. Attention service regressions cover pending question forms and confirmations in persistent agent chats, removal after an answer with retained history, and company/task-visibility/addressee checks. Idle chats without structured requests remain absent. The Needs You component test checks that a chat-sourced canonical question reaches its original resolver and context. It also checks simultaneous resize/light and delayed movement to Recent Decisions. Shared card-motion tests cover reduced-motion settlement and returning to a resolved card without replay. The decision-motion stories exercise real Needs You/resolver components with isolated approval, question, and failed-save responses. Projection helpers cover distinct pending task membership and presentation-only recency decay. Browser verification must additionally approve copy, approve a visual, and answer a question; confirm three retained receipts, task/chat consistency, resize/collapse motion, and exact task context on receipt click. Turning the setting off restores the classic surface.
search
Automated: search page and company search service cover UI and server search separately.
Manual: Search a unique task title and content phrase, open the result, and compare the actual record. Test filters, no results, and a second company. Verify stale results do not expose inaccessible data.
toasts
Automated: persisted eligibility
checks creation, assignment, comments, mentions, deliberate edits, excluded read
activity, company isolation, local-board, and persisted run-task attribution.
Socket delivery checks
viewer-specific projection; UI delivery
checks eligibility gating, cache updates, stale callbacks, and existing suppression.
UI cache handling checks visible
comment refresh when authenticated events omit hydration metadata.
The task toasts follow personal involvement in multiplayer case in the
authenticated browser suite
exercises quiet unrelated updates, read-only visits, first/historical mentions,
self-activity, and visible-task suppression against persisted data.
Manual: Use two authenticated browser sessions in the same isolated company. Create separate tasks for each person and use one shared agent. From a company task list, verify that the other person's unrelated task updates produce no toast but still refresh the list. Comment on one of their tasks, then have them update it: a toast should appear. Repeat using assignment, a deliberate edit, and a first explicit user mention in a comment or description. Reload before a later update to prove eligibility does not depend on browser history. Mere viewing, agent ownership, responsible-user attribution, and parent participation must remain quiet. Task-linked failures follow the same participation rule; generic agent errors remain quiet. Repeat while viewing the relevant task and after switching company/account to verify suppression and isolation.
Gotchas
-
Command-palette and keyboard entry points also reach search. Verify quick-filter/query handoff separately from opening the full search route.
-
Notifications do not grant task access or change Mine membership.
-
A feature-map recipe and mocked UI tests do not prove live multiplayer behavior.
-
Combined Inbox/Tasks requires both
enableCombinedInboxTasksandenableStreamlinedUi. Standalone Inbox tests render archive/read controls and toolbars; this is not live multi-user evidence. -
Personal archival is not task cancellation or completion.
-
Legacy Inbox routes can redirect to the combined task view; record which host actually rendered.
-
Search service tests do not establish the relevance quality of every query.
Storybook verification
views
Capability ID: inbox-search/views.
Scope: ui Storybook: Mine/Recent/Unread/Blocked and broader task views; combined versus standalone host; loading/empty/error/filter states. Production: Use persisted personal participation/read data; verify scoped counts and list navigation/reload; automatic read cleanup must not create Mine membership. Sources: Source: Inbox.tsx, Issues.tsx. Tests: Inbox.test.tsx, Issues.test.tsx, issues-service.test.ts
toasts
Capability ID: inbox-search/toasts.
Scope: ui Storybook: Eligible live task toast, explicit mention, quiet unrelated event, own-action/visible-task suppression, dedupe and company/account switch. Production: Persisted participation and mentions determine per-viewer socket eligibility; quiet events still refresh caches; reload retains eligibility, unauthorized resources remain denied. Sources: Source: ToastViewport.tsx, ToastContext.tsx, LiveUpdatesProvider.tsx, task-toast-eligibility.ts. Tests: ToastContext.test.tsx, ToastViewport.test.tsx, LiveUpdatesProvider.hook.test.tsx, task-toast-eligibility.test.ts, live-events-toast.test.ts, multi-user-authenticated.spec.ts
Coordinator navigation counts all outstanding canonical attention items on desktop and phone, using the same complete feed as Needs You. Classic keeps its existing today-oriented badge. Task questions resolve through their original interaction; pending chat projections, counts and recorded receipts invalidate together.