mirror of
https://github.com/nexu-io/open-design.git
synced 2026-10-27 09:51:12 +01:00
a019bcd5ccf1a6f7a181a8f518b1654ce248fb0b
1368
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
a019bcd5cc |
fix(web): backport release-only fixes to main ahead of v0.10.0 re-cut (batch 1) (#3711)
* fix(web): avoid duplicate style picker group keys (cherry picked from commit |
||
|
|
9a25aae339 |
feat(landing-page): add Solution hub at /solutions/ and consolidate Role pages (#3700)
* refactor(landing-page): move Solution → Role pages from /for/ to /roles/ The Solution → Roles pages shipped at /for/<role>/ in #3658. The /for/ prefix reads as an incomplete URL fragment on its own; /roles/<role>/ is a clearer, self-describing path for the audience landing pages. Rename both the default-locale and [locale] page directories, update the header dropdown hrefs and the route docstrings to match. R2 hero image keys stay under landing/assets/for/<role>/ (the bucket key is not user-visible and the assets already exist there), so only the public route changes. Old /for/<role>/ URLs are dropped, not redirected — the pages are hours old with negligible inbound links. * feat(landing-page): add Solution hub at /solutions/ and fold Role pages in Builds on the Solution → Role pages from #3658. Two changes to the Solution information architecture: 1. The Solution nav trigger pointed at /compare/ (the competitor comparison hub), which is unrelated to the dropdown's contents. It now points at a new /solutions/ overview page that indexes every Solution sub-page, grouped exactly like the dropdown: Use cases (what you're making) and Roles (who you are). Each card reuses the sub-page's own breadcrumb + description from getSolutionPageCopy, so the hub never drifts from the pages it links to. The page's own chrome (label, heading, one-line lead) lives in a new solutions-index-i18n module, typed as a Record over every landing locale so a missing translation is a typecheck error; all 18 locales are populated. 2. The Role pages, which shipped at /roles/<role>/, move under /solutions/<role>/ so they sit alongside the use-case pages in one coherent /solutions/ namespace instead of a parallel /roles/ tree. Use-case and role slugs don't collide. Dropdown hrefs and route docstrings updated to match. The interim /roles/ paths are dropped, not redirected (hours old, negligible inbound links). Also: the brand logo's default href was '#top', leaving '/#top' in the URL on the homepage and pointing at a non-existent anchor. It now defaults to '/' so clicking the logo always returns to the localized home, matching what every sub-page caller already passed. Verified: landing-page typecheck (0 errors) and build:static — output has /solutions/ + all 11 sub-pages across 18 locales, the /solutions/ index, no /roles/ or /for/ routes remain, and hero images still resolve to their existing R2 keys. --------- Co-authored-by: Joey-nexu <joeylee12629@gmail.com> |
||
|
|
f1b59b67c6 |
feat(web): streamline Home/project composer chrome and pin the Home tab (#3692)
* feat(web): streamline Home/project composer chrome and pin the Home tab Reduce composer clutter and align the Home and project chat surfaces: - Drop the artifact-specific footer settings (fidelity/ratio/notes/etc.) from the Home composer; the agent now asks for them via the existing discovery flow, so the footer keeps only the design-system picker. - Default the design-system picker to "Auto"; only a user-owned system pre-selects. - Move the CLI (model switcher) into the top-right header next to Discord; unify the composer footer buttons to pill shape. - Extract a shared ComposerPlusMenu (attach / connectors / plugins / MCP, with an extra design-toolbox flyout in the project composer) used by both Home and the project chat composer; remove the project composer's sliders and @ buttons (mentions still trigger by typing @). - Rename the session-mode label to "Design" (full "Design Agent" copy stays in the hover card). - Pin the Home workspace tab to the leftmost slot, make it non-closable, and tighten tab widths. - Fix the truncated Discord online count in the icon-only top bar; ellipsize long plugin/MCP names in the add menu. * fix(web): address composer + menu review regressions and update tests - ComposerPlusMenu: reset the shared search query when the active submenu changes, so a plugin search no longer cross-filters the MCP flyout (restores the old HomeHero behavior). - ChatComposer: opening the "+" menu latches composerEngaged again, so the Plugins / Connectors / MCP submenus aren't empty on a cold composer. - Keep media-surface footer options (model / ratio / resolution / duration / audio type); only prototype/deck fidelity/slide-count/speaker-notes were meant to leave the footer. - Update/trim tests for the removed sliders + tools popover, the @ button, the Import-menu design-system switcher, and the prototype/deck artifact footer controls; rewire the design-toolbox tests through the new "+" menu; assert the Home tab is pinned and non-closable. * fix(web): guard composer + menu picker mousedown to preserve caret The shared ComposerPlusMenu connector/plugin/MCP picker rows ran onPick* handlers (which call insertMention) without cancelling mousedown, so the button stole focus first and Lexical fell back to selectEnd(), inserting mentions at the end of the draft instead of the caret. Add the preventDefault guard and port the deleted caret-regression coverage to the shared menu path. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * fix(web): defer prototype/deck artifact inputs and protect + menu caret Address the two blocking review findings on PR #3692: - HomeView.submit() now strips fidelity / slideCount / speakerNotes from the submitted plugin inputs via stripArtifactFooterInputs(), so removing the footer actually defers those questions to first-turn discovery instead of forwarding the plugin manifest defaults. The prose brief still renders them. - ComposerPlusMenu pick rows (connector / plugin / MCP) cancel mousedown so the editor keeps focus and the caller's insertMention lands at the caret, not the draft end — restoring the #3195 contract that the deleted tools-menu-caret test used to cover. - Add ComposerPlusMenu.test.tsx covering the mousedown-preventDefault contract and the submenu query reset; update prefill expectations to assert the artifact fields are no longer in the apply payload. * fix(web): keep Home tab pinned leftmost on tab drop Re-normalize reordered tabs and coerce 'before Home' drop edges to 'after Home' so dragging a project tab onto Home's left edge can no longer persist a tab ahead of the permanent Home tab. Updates the drag spec to assert Home stays first after the drop. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * fix(web): always mint a pinned Home tab during tab normalization normalizeTabsState only reordered an existing Home tab, so users who closed/replaced Home before this feature shipped could restore a saved [project, ...] workspace and stay Home-less until manually navigating home. Prepend a Home tab whenever none survives normalization so the "Home always exists and is leftmost" invariant also holds for migrated state. Adds a regression spec booting from saved tabs with no Home entry and updates two specs that encoded the old Home-less behavior. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) --------- Co-authored-by: qiongyu1999 <2694684348@qq.com> |
||
|
|
8f6cdaaba1 |
Harden release-beta-s self-hosted beta lane (#3641)
* Fix Windows smoke launcher payload layout
* Harden release-beta-s runner probe
* Use cmd shell for release-beta-s probe
* Add win beta build script for self-hosted lane
* Keep release-beta-s checkout lightweight
* Use explicit local beta version for release-beta-s
* Install dependencies in beta build script
* Repair Electron dist before Windows beta build
* Add release-beta-s timings and diagnostics
* Polish release-beta-s observability guardrails
* Trace tools-pack cache materialization
* Reuse materialized workspace build cache
* Trace Windows packaging segments
* Trace Windows archive process details
* Reuse installer unpacked materialization
* Preserve reusable Windows unpacked projection
* Add fast smoke telemetry for self-hosted beta
* Cache Windows archive outputs for beta lane
* Tighten fast Windows smoke lane
* Scope Windows beta artifact reporting by target
* Clean stale Windows target artifacts
* Wire Windows Authenticode signing
* Expose self-hosted beta validation modes
* Add core Windows smoke profile
* Decouple core smoke from updater fixture version
* Trace Windows smoke lifecycle steps
* Skip unpacked materialization on NSIS cache hit
* Add Nexu S3 beta publisher
* Harden self-hosted Windows beta lane
* Use Windows PowerShell for publish probe
* Run publish probe via repo script
* Require explicit self-hosted S3 public origin
* Run real installer acceptance for external feeds
* Skip ready-prompt checks for external feeds
* Relax optional self-hosted update inputs
* Tolerate BOM in self-hosted publish probe
* Converge self-hosted beta metadata and signing flow
* Run self-hosted beta metadata prep on the runner lane
* Use Windows PowerShell for self-hosted beta metadata checks
* Bypass execution policy in self-hosted beta metadata checks
* Split self-hosted beta metadata publish
* Suppress mc output in beta publish helpers
* Split Windows NSIS payload cache layers
* Require fresh tool dist metadata
* Preserve Windows overlay payload paths
* Default beta-s gate to core
* Optimize win beta packaging reuse
* simplify windows full smoke lifecycle
* preserve onboarding state after windows upgrade smoke
* fix windows tools-pack test portability
* recover stale tools-pack locks
* Allow auto signing fallback for beta builds
* Add self-hosted beta mac arm64 lane
* Quote self-hosted signing choices
* Run self-hosted beta metadata on mac lane
* Use shallow metadata checkout on self-hosted beta
* Use sparse beta metadata checkout
* Make mac release profile bash-compatible
* Allow file sparse metadata checkout
* Isolate self-hosted metadata checkout
* Clear sparse skip-worktree before mac build
* Isolate self-hosted mac build checkout
* Run self-hosted beta publish without aws cli
* Use unsigned beta prefix for mixed platform publish
* Sparse checkout self-hosted beta publish
* Publish mac beta platform before metadata
* Pass mac publish release profile
* Probe mac signing keychain import
* Expand mac signing preflight variants
* Use sudo keychain helper for self-hosted mac signing
* Route self-hosted mac codesign through sudo wrapper
* Normalize mac signing identity name
* Harden self-hosted mac notarization
* Optimize self-hosted mac release cache
* Fix mac signing helper cleanup
* Retry transient mac notarization uploads
* Keep mac notarization fail-fast
* Reduce mac symlink diagnostics noise
* Normalize self-hosted beta platform inputs
* chore: bump beta base version to 0.9.1
* chore: tune release tools-pack cache retention
* test: add temporary cache validation marker
* chore: key windows nsis base payload by content
* Revert "test: add temporary cache validation marker"
This reverts commit
|
||
|
|
d0d9c02763 |
feat(web): Manus-style second-level category rail on Home composer (#3686)
* feat(web): add Manus-style second-level category rail to Home composer Picking Prototype or Slide deck on Home now reveals a compact sub-category pill row (All + the Community facet sub-categories), filtering the example preset cards in place without binding a plugin or stamping a badge. Reuses the existing plugins-home facet taxonomy so the labels match the Community grid exactly; Landing / marketing is floated first for prototypes. Also: reorder the first-level rail (Prototype, Slide deck, HyperFrames, Live artifact, Image, Video, Audio); move the CLI/BYOK execution switcher out of the composer footer up to the top bar beside Discord; restyle the example preset cards into full-bleed thumbnail + name (dropping the prompt blurb, trailing icon, and faux browser chrome); and stop showing the "Example prompts: …" badge when a preset is applied. Co-Authored-By: Claude Opus 4 <noreply@anthropic.com> * fix(web): derive Home sub-category pills from the displayed preset list Addresses review: a selected child pill could render every preset card instead of its slice. activeSubChips was built from the full install set (pluginOptions) while filtering ran against the curated/sorted/sliced activeExamplePlugins, so a facet category whose plugin wasn't in the Home preset slice produced a pill that, when clicked, hit the all-cards fallback and looked unfiltered. Derive the pills from activeExamplePlugins (the same list that feeds the cards) and drop the fallback, so every pill maps to a non-empty slice. Update the rail test (preset card is now thumbnail + name, no prompt blurb on the face) and add a fixture for a sub-category absent from the displayed list. Co-Authored-By: Claude Opus 4 <noreply@anthropic.com> --------- Co-authored-by: qiongyu1999 <2694684348@qq.com> Co-authored-by: Claude Opus 4 <noreply@anthropic.com> |
||
|
lefarcenandGitHub
|
4cfdc4d7db |
feat(desktop): play white brand splash video on startup (#3674)
* feat(desktop): play white brand splash video on startup
Replace the dark "Waiting for the web runtime URL…" placeholder with the
brand intro animation on a white background. The clip plays once and then
freezes on the settled "Open design" logo. The window only swaps to the web
runtime after the clip has had time to play through (MIN_SPLASH_MS), so it
never flips mid-animation, and it holds the final frame when the runtime
takes longer to boot.
The clip is embedded as a base64 data URL so it renders identically in dev
and in packaged builds without any electron-builder asset wiring — the
packaged app exposes no reliable on-disk asset path from the desktop main
bundle (the previous icon lookup already silently fell back to null there).
* feat(desktop): show splash in parallel with boot, reveal on web mount
Make the white brand splash a dedicated window so it can do real work:
- Packaged cold start now creates the splash BEFORE awaiting the daemon/web
sidecars, so the animation is on screen during the whole boot instead of the
user staring at no window at all for a few seconds. The 2s minimum splash now
overlaps startup rather than being added on top of it.
- The main window starts hidden and loads the web app behind the splash. We
reveal it only once the web bundle reports it has actually mounted
(data-od-app-mounted="1") and the minimum splash time has elapsed, then close
the splash. This removes the dark "Loading Open Design…" web shell that used
to flash between the splash and the app. A hard timeout guarantees the user is
never stranded on the splash if the mount signal never arrives.
tools-dev/dev keeps working unchanged: when no splash window is handed in, the
runtime creates its own.
* fix(desktop): measure splash hold from window creation, not adoption
PerishCode review: in packaged startup the splash window is created before
`await startPackagedSidecars(...)`, but the minimum-hold clock was stamped later
inside `createDesktopRuntime` (after the sidecars had finished booting). The
reveal then waited `MIN_SPLASH_MS` from that post-boot instant, so a fast web
mount still added almost another 2s on top of boot — re-introducing the very
delay the parallel splash was meant to remove.
`createSplashWindow()` now returns a `{ window, startedAt }` handle stamped at
creation; the packaged entry forwards `startedAt` via `splashStartedAt` and the
runtime measures the hold from there. Runtime-created dev splashes stamp their
own creation time, so the hold overlaps boot in every path.
* fix(desktop): gate DesktopRuntime.show() behind the splash reveal
PerishCode review: the main window now starts hidden and is revealed only after
the web app mounts, but `DesktopRuntime.show()` still called `window.show()`
unconditionally. The packaged entry routes a pending second-instance focus
through that method via `onDesktopReady`, so launching the app twice while the
sidecars/web are still booting would surface the half-loaded main window and
bypass the reveal gate — reintroducing the startup flash.
`show()` now focuses the splash while the runtime is pre-reveal and only shows
the main window once `revealWhenReady()` has fired.
|
||
|
|
475b9a480b |
fix(daemon): include user toolchain dirs in agent spawn PATH so shebang interpreters resolve (#3675)
What: `applyAgentLaunchEnv` now appends the user toolchain bin dirs
(Homebrew, ~/.local/bin, ~/.bun/bin, version-manager node dirs, npm
prefixes — the same set binary *resolution* searches) to the end of the
spawn PATH. Exposed via a new `userToolchainBinDirs()` from executables.ts;
injectable param defaults to the real dirs (tests pass []).
Why: agent detection resolved a binary's path using resolvePathDirs()
(PATH + wellKnownUserToolchainBins) but spawned it with a PATH containing
only nodeBinDir + the agent's own dir. An agent whose shebang interpreter
lives in a toolchain dir not on the daemon PATH then failed to execute:
Pi (`/usr/bin/env bun` via @oh-my-pi/pi-coding-agent) resolved fine but its
`--version` probe exited 127 ("env: bun: No such file or directory"), so
detection classified it not-invocable and marked Pi unavailable. claude/codex
only survived because their interpreter (node) happened to be on the spawn
PATH. Resolution and spawn PATH are now symmetric, fixing detection AND the
run/connection-test spawns that share this function.
How: append appendPathDirs into the PATH merge (lowest priority, deduped).
Verified: Pi flips unavailable -> available after the fix (path + version
now populated), alongside claude/codex.
Test: apps/daemon/tests/runtimes/launch.test.ts — "appends toolchain dirs
so shebang interpreters (e.g. bun) resolve at spawn time" (would go red
without the fix). Existing exact-PATH tests pass [] to stay deterministic.
Co-authored-by: ZC Deng <zhic.deng@gmail.com>
|
||
|
lefarcenandGitHub
|
9fd65efec0 |
fix(web): scope generation-preview override to the empty design-files tab (#3684)
Address review: dropping the DESIGN_FILES_TAB guard outright let the generation card win for every in-flight run on the design-files tab, hiding the file browser of a populated project until a previewable artifact existed. Scope the override to the truly empty tab, mirroring DesignFilesPanel's own empty-state gate (no files, no live artifacts, no folders), so a populated project keeps browsing while generating. Adds a regression test that a populated design-files tab still renders file rows while a run is active. |
||
|
|
1b47dcd22c |
feat(landing-page): add Solution → Roles pages (/for/<role>) for 5 audiences (#3658)
* feat(landing-page): add Solution → Roles pages (/for/<role>) for 5 audiences Enable the Solution dropdown's Roles group (previously commented out) and ship the five audience landing pages it points to: solo builders, designers, engineers, product managers, and marketing teams. Why: the Solution nav shipped with use-case pages but the Roles group was held back with no destinations. Visitors arriving by job-to-be-done had no "this is for someone like me" entry point, and the use-case pages alone don't speak to a role's full workflow. Each role page is a conversion surface (role search terms have ~zero volume, so these are not SEO-traffic pages) and an internal-link anchor for the Solution cluster. What users see: a Roles group at the bottom of the Solution dropdown with five links. Each page has its own hero, a role-specific TL;DR, a three-step "how a <role> uses Open Design" flow, a feature grid of real catalogue templates curated per role (no overlap between roles — every role shows a distinct set of interactive examples), a role-specific comparison table, and a role-specific FAQ. All copy is translated across the 18 landing locales; the nav group label is localized too. Pages reuse sub-page-layout and emit BreadcrumbList + WebPage + FAQPage JSON-LD with locale-aware canonical URLs. Featured templates are picked by manifest id through pickSolutionPlugins, which fails the build on any missing id, so a renamed or removed example breaks the build loudly instead of shipping a short grid. * fix(landing-page): align nav dropdown caret to the label's optical midline The ▾ caret after each top-level nav label (Product, Solution, Agent, …) rendered as a baseline-aligned inline glyph at 10px, so it dropped to the lower-right of the label and read like a stray comma after each word — especially visible with CJK labels. Make the nav link an inline-flex row centered vertically, bump the caret to 11px with line-height 1, and nudge it up 1px to sit on the text's optical midline. The hover/focus state keeps the same resting position (the color shift to coral remains the open affordance) so the glyph no longer jumps. * refactor(landing-page): split solution-pages-i18n into per-locale modules The Roles pages added five role* keys with full prose across all 18 locales, which grew solution-pages-i18n.ts to ~1.4 MB — over the repo's 1 MiB MAX_CHANGED_FILE_BYTES blob guard (.github/workflows/ci.yml), making the required "Guard changed file sizes" check (and the Validate workspace aggregate that depends on it) red and the PR unmergeable. Move each locale's copy block into solution-pages-i18n/<code>.ts and the shared shapes into solution-pages-i18n/types.ts. solution-pages-i18n.ts is now a thin index: it re-exports the public type API, imports the per-locale modules, assembles BY_LOCALE, and resolves copy with the English fallback. The public API (getSolutionPageCopy, SolutionPageKey, SolutionPageCopy, …) and runtime behavior are unchanged. Largest split file is ~110 KB (ru), the index is ~3 KB, so every tracked source file is now well under the 1 MiB limit. Verified: typecheck clean, build green (8457 pages), and per-locale role-page titles still render translated (zh/ja/de/ru/ar) rather than falling back to EN. --------- Co-authored-by: Joey-nexu <joeylee12629@gmail.com> |
||
|
|
0767948f48 |
fix(web): tighten preview workspace panels (#3643)
* fix(web): tighten preview workspace panels * test(web): align preview panel assertions Generated-By: looper 0.9.2 (runner=fixer, agent=codex) * fix(web): simplify handoff cli unavailable rows * fix(web): clarify handoff framework selection * fix(web): refine preview comment controls * fix(web): align queued send strip with composer * fix(web): balance mark toolbar wrapping * fix(web): let mark note actions shrink --------- Co-authored-by: chaoxiaoche <chaoxiaoche@chaoxiaochedeMacBook-Pro.local> Co-authored-by: Looper <looper@noreply.github.com> |
||
|
|
8920577f43 |
feat(web): simplify Home composer (drop plugin-inputs form & example pill, move Design Agent next to Send) (#3645)
* feat(web): simplify Home composer (drop plugin-inputs form & example pill, move Design Agent next to Send) - Remove the inline plugin-inputs form (ARTIFACT KIND / AUDIENCE / TEMPLATE) from the Home composer; required inputs no longer gate Send (defaults still flow to the backend). - Remove the dismissible "selected example" pill above the editor; picking an example still seeds the prompt. - Move the Design Agent toggle from the footer-left group to sit directly left of the Send button. * fix(web): keep client-side required-input gate on Home composer Dropping the inline plugin-inputs form left pluginInputsAreValid() always returning true, so a required field with no default that the user blanks out in an editable Home scenario kept Send enabled and only failed at click-time apply with a generic "Failed to apply …" error (the daemon's validateInputs rejects missing required inputs). Restore a client-side gate that mirrors the daemon's required-input rule and names the missing field in the submit error, so these Home flows fail fast. Extracts the pure check into utils/pluginRequiredInputs with unit coverage. * fix(web): drop example from active-row mount condition With the example pill removed from the Home composer active row, keeping selectedPromptExample in showActiveContextRow mounted an empty .home-hero__active row when an example was picked with no other active context. Remove it from the condition so the row only renders when it has visible content. --------- Co-authored-by: qiongyu1999 <2694684348@qq.com> Co-authored-by: lefarcen <935902669@qq.com> |
||
|
lefarcenandGitHub
|
9a18397090 |
fix(web): keep generation progress card on the design-files tab (#3671)
The right-side preview pane showed the empty "Creations will appear here" file list while an agent was actively generating, instead of the generation progress card with its understand/generate/prepare steps. PR #3516 (Lexical composer rewrite) added an `activeTab !== DESIGN_FILES_TAB` clause to `showGenerationPreview`. The design-files tab is the default landing tab, so on a fresh project the in-flight generation preview was suppressed and the idle empty state won the surface. Before #3516 the preview branch rendered ahead of the design-files branch with no tab guard. Drop the clause so the progress card again takes priority over the empty file list. `buildGenerationPreviewState` returns null once a previewable artifact exists, so a finished artifact is never hidden behind the card. Adds a red-spec FileWorkspace test that renders a generating run on the design-files tab and asserts the generation-preview stage is shown. |
||
|
lefarcenandGitHub
|
e19c057d8d |
fix(web): flip deck preview to a queued send's marked slide (#3668)
* fix(web): flip deck preview to a queued send's marked slide When several edits are queued against marked elements on different deck slides, the preview stayed parked on slide 1 while the agent drained the queue — the user couldn't see each edit land in context. The mark's slide index was already captured into the comment snapshot and target, but it was dropped when the snapshot became a ChatCommentAttachment, so nothing downstream knew which slide a queued send touched. Carry `slideIndex` through the contract and both attachment builders, then have ProjectView arm a `slideNavRequest` (mirroring `shareRequest`) the moment a queued send starts processing. FileWorkspace activates the deck tab if it is open and FileViewer consumes the nonce once to flip the preview via the existing `od:slide` go bridge, writing the cached slide state first so a freshly-mounted iframe restores to the same slide on load. No-op for plain prompts, free pins, or marks without a slide index. * fix(web): keep deck slide-nav consume-once across viewer remounts The consumed-nonce guard lived in a per-mount ref, so leaving the deck tab and returning remounted HtmlViewer, reset the ref, and replayed the still-live slideNavRequest — yanking the preview back to the queued slide and clobbering manual navigation. Move the guard into shouldConsumeSlideNav, keyed by preview state key outside the component, so "consume once" survives remounts. Add a unit spec covering first-handling, fresh-nonce, and per-file isolation. * fix(web): only flip a deck whose tab was open when the queued send started slideNavRequest stayed live in parent state, so a request for a deck that was closed at processing-start would resurface and jump the preview when the user opened that file by hand much later — not the follow-along behavior the feature promises. Decide deliverability once, at fire time: FileWorkspace marks the nonce deliverable only when the target deck is already an open tab, and the viewer is fed the request only for that deliverable nonce on the active file. A closed deck never becomes deliverable, so opening it later cannot jump. Pure helpers isSlideNavDeliverableNow / deliverableSlideNavForActiveFile carry the decision, with specs for the closed-deck-opened-later case. |
||
|
netteeandGitHub
|
c6ee16be9d | feat(daemon): pass open design trace env to AMR (#3667) | ||
|
|
aa782365e0 |
feat(daemon): pass parentSession in pi RPC new_session for conversational continuity (#3434)
* feat(daemon): pass parentSession in pi RPC new_session for conversational continuity
Add parentSession option to attachPiRpcSession that sends
new_session { parentSession } before prompt on pi's stdin.
pi loads the prior .jsonl session file, forking conversation
history into the new session.
- resolveLastSessionPath(cwd) scans <cwd>/.pi/sessions/ for
the most recently modified .jsonl
- piParentSessionPaths Map in server.ts persists the last
session path per effectiveCwd
- Backward compatible — no new_session sent when no prior
session exists
- 3 new tests covering stdin ordering, backward compat, and
session path capture after agent_end
Closes #3414
* fix: replace str shorthand with string in parentSession test type annotations
* fix(daemon): scope pi session resume to conversations
* fix(daemon): handle pi parent session failures
* fix(daemon): clear stale pi session on capture miss
* fix(daemon): keep pi resume failure SSE code stable
---------
Co-authored-by: Siri-Ray <2667192167@qq.com>
|
||
|
lefarcenandGitHub
|
e260b56b30 |
fix(web): bottom-anchor the chat transcript so short and forked conversations don't strand at the top (#3669)
A short conversation — most visibly a fresh "Fork from here" branch — rendered its messages stuck to the top of the chat log with a large empty area below, instead of resting against the composer like a normal chat. Two causes: 1. `.chat-log` is a flex column with no vertical anchoring, so when the content is shorter than the viewport the messages sit at the top (~530px of dead space below). 2. The anchor-to-top reserve (`.chat-log-tail-spacer`, sized while a reply streams so the user's turn can reach the top) never collapsed once the turn finished. A short completed turn was left pinned at the top above ~540px of reserved blank. Fix: - Bottom-anchor the transcript with `margin-top: auto` on the first row. When content is shorter than the viewport the auto margin absorbs the slack and the messages rest against the composer; when it outgrows the viewport the margin resolves to 0 and the log scrolls normally with the top fully reachable. (Deliberately not `justify-content: flex-end`, which clips overflowing content above the scroll range.) - Collapse the tail spacer when streaming ends, and reset it (plus the anchor flags) on conversation switch, so a finished or freshly opened short conversation settles cleanly instead of inheriting a stale reserve. Anchor-to-top during an active turn is unchanged. Verified live (Playwright, local daemon+web): fresh short conversation last-message-to- bottom gap 532px -> 20px; completed short turn tail spacer 549px -> 0; long conversation still scrolls and reaches the top (scrollTop 0, first row at +18px). |
||
|
lefarcenandGitHub
|
b33ea26de3 |
fix(web): stop conversation list flashing "0 msg" while switching chats (#3663)
* fix(web): stop conversation list flashing "0 msg" while switching chats The chat tab's conversation list derives the active conversation's message count from the live `messages` array. Switching conversations resets that array to [] while the new conversation's history is fetched, so the active row rendered a phantom "0 msg" for the duration of the load (and stayed at 0 if the load failed) instead of its real count. Trust `messages.length` for the active conversation only once the live array actually reflects it (`messagesConversationId === activeConversationId`); otherwise fall back to the persisted server `messageCount`. The live count is still used during streaming so the active row stays fresh as turns arrive. * fix(web): wire messagesConversationId for the other ChatPane mounts SideChatTab and DesignSystemFlow also render the conversation list but did not pass the new marker, so their active row fell back to the persisted count and lost live-count freshness. Both own a single loaded conversation whose messages always reflect the active id, so pass that id through to keep the active row's live count authoritative (restores pre-change behavior). ProjectView still passes its real messagesConversationId (null during the switch/load window). * fix(web): keep secondary ChatPane mounts on the persisted message count The SideChatTab and DesignSystemFlow mounts both have loader windows where the live `messages` array does not reflect the active conversation — useConversationChat resets to [] before a load, and DesignSystemFlow's loader does not retag projectChatMessages on switch. Marking the live array as authoritative there would reintroduce the same phantom/stale count this PR removes. Omit messagesConversationId for those mounts so their list uses the stable persisted conversation.messageCount; only ProjectView, which tracks the marker precisely, opts into the live count. |
||
|
|
7ab1e6eff2 |
feat(composer): reconcile context with prompt + per-connector brand color (#3625)
* feat(composer): reconcile context with prompt + per-connector brand color Two increments on top of the Lexical composer (#3516): - Filter contextPlugins/contextMcpServers/contextConnectors at submit time by whether the entity's @mention still survives in the prompt text. When a user deletes a mention pill, that plugin/MCP/connector is no longer sent to the agent. Adds mentionTokenPresent() to inlineMentions. - Give connector mention pills a per-connector brand color (Notion black, Figma purple, etc.) by setting an inline --m-hue in MentionNode.createDOM, reusing the existing CSS hue machinery instead of the shared connector green. Adds a curated brand-color map with a hashed fallback. * fix(composer): keep mention context when punctuation trails the pill mentionTokenPresent() reused the parser's strict isMentionRightBoundary(), so a still-visible atomic pill serialized as `@Slack,` was treated as deleted and dropped from contextPlugins/contextMcpServers/contextConnectors at submit time. Add a submit-time boundary that also accepts trailing clause/sentence punctuation, leaving the parser invariant untouched, plus a HomeView regression for the punctuation case. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * fix(composer): make connector pill brand hue theme-aware Near-black curated brand colors (Notion, GitHub, Slack) rendered the connector mention pill text as dark-on-dark in the app's dark theme, since chat.css derives the text as color-mix(--m-hue 72%, --text) and --text is light in dark mode. connectorBrandColor now takes the live theme and lightens hues below a luminance floor so they stay legible, while bright brands (Figma, Linear) pass through unchanged. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * fix(composer): make connector pill hue follow live theme switches Recompute a mounted connector pill's brand --m-hue when <html data-theme> or the OS prefers-color-scheme changes, not only on node create/update, so a pill inserted in light mode no longer keeps its near-black hue (and dark-on-dark text) after a live switch to dark mode. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * feat(composer): collapse footer tools into one "+" menu; drop duplicate context badges - Replace the separate "@" context-trigger and paperclip attach buttons with a single "+" button. Its popup offers Attach files plus Connectors/Plugins/MCP right-side flyout submenus (Plugins/MCP searchable + scrollable). Picking an entry inserts the inline @mention pill via the existing pickX handlers; "Add …" rows route to the plugins / integrations views. - Stop rendering the top plugin/connector/MCP context-badge row — those now live as inline mention pills in the composer, so the badges were duplicates. Staged file chips still render there. showActiveContextRow no longer keys off the selected*Contexts arrays. - New i18n keys (addMenu/addPlugin/addConnectors/addMcp/noPlugins/noMcp/ noConnectors) in types.ts + en + zh-CN; other locales inherit via ...en. * fix(composer): repair collapsed-menu test selector and complete addMenu i18n Update the HomeView context-picker test to open the plus menu before asserting the attach affordance, since the collapsed footer now renders home-hero-plus-attach only after home-hero-plus-trigger is clicked. Add the seven home-hero add-menu keys (addMenu, addPlugin, addConnectors, addMcp, noPlugins, noMcp, noConnectors) to every locale beyond en/zh-CN so the i18n contract holds across all 18 locale files. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * test(web): drop assertions on removed context-badge row The PR removed the duplicate plugin/connector/MCP context-badge row (context now renders as inline @mention pills), but four HomeView tests still asserted the dropped `home-hero-context-plugin-*` testids and went red. Re-anchor them to behavior that survives: the submit payload's `contextPlugins`/prompt-pill text for picker flows, and the active context row's resolved-count label for empty-prompt "Use" handoffs. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * test(web): align entry-chrome e2e with collapsed footer menu The footer-collapse change removed the always-visible home-hero-attach control and the plugin/MCP/connector context-badge row, but the entry chrome Playwright suite still asserted those removed surfaces, failing Browser tests (and the Validate workspace aggregate). - Assert home-hero-plus-trigger instead of the removed home-hero-attach. - Drop the home-hero-context-plugin-* badge assertions; plugin context no longer renders a visible badge. Coverage is preserved by the existing prompt-text and submit-payload (pluginId) assertions. * fix(composer): exclude inline-mentioned contexts from active row; reset plus-menu query per submenu Gate the Home active context row on contexts that have no inline @mention representation, so selecting only an inline-mentioned connector no longer mounts an empty row above the editor. Reset the shared plus-menu search query when the active submenu changes so a stale plugin search no longer filters the MCP list into a false "No MCP servers" empty state. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * fix(composer): keep context-only Use selections in submit payload The submit-time reconciliation dropped every selected context that lacked an inline @mention token, which silently discarded contexts staged through the plain Use action (it never writes a token). The agent then received no contextPlugins even though the user explicitly chose the plugin as context. Tag each selection with inlineBacked: contexts inserted as @mention pills are only sent while their token survives in the prompt; context-only Use selections stay in the payload until explicitly cleared. The active context row likewise counts only context-only selections, so an inline-mentioned-only pick no longer mounts an empty row. Adds a regression covering the direct Use submit path. * fix(composer): render removable chips for context-only Use selections Context-only selections (staged via the plain Use action, no inline @mention pill) are kept in the submit payload, but the home composer no longer rendered any chip for them, so they were invisible and unremovable — a kept context would silently drift into later freeform prompts. Render a removable chip in the active context row for each context-only plugin/MCP/connector selection, wired to the existing remove handlers. Adds a regression that stages a plain Use context, clears it via the chip, and asserts the submit payload no longer includes it. --------- Co-authored-by: qiongyu1999 <2694684348@qq.com> Co-authored-by: lefarcen <935902669@qq.com> |
||
|
lefarcenandGitHub
|
e547ba28a4 |
fix(web): stop early analytics events shipping app_version=0.0.0 (#3609)
useAppVersion() reads the daemon-pinned version from /api/version at runtime, but PostHog's app_version rode solely on the register() super-property, which races that fetch. High-frequency early events (page_view above all) captured before the version resolved shipped with app_version='0.0.0' permanently — the dashboards showed 0.0.0 on every build after the runtime-fetch approach landed. resolveAppVersionForCapture() awaits a shared single-flight /api/version fetch before every capture path (client init, super-property register, track, consent re-init) and stamps app_version/ui_version onto each event explicitly, so no event ships with the placeholder once the version is knowable. The hook still returns the placeholder for first paint. Red spec: tests/analytics-app-version.test.tsx gains resolveAppVersionForCapture cases (resolves real version before capture; no refetch when already real). |
||
|
|
234a11aa05 |
feat(web): collapsible entry rail + quiet new-user templates + icon-only topbar chips (#3617)
* feat(web): collapsible entry rail + quiet new-user templates + icon-only topbar chips Rework the entry view to keep new users focused on the hero composer: - Nav rail collapses to a single topbar toggle by default and docks as an overlay on demand; a hover-reveal collapse button on the logo folds it back. - For users with no projects, the Community templates gallery starts hidden behind a faint bottom hint; an upward scroll gesture or click smoothly reveals it, with a pinned collapse pill to fold it away again. - Collapse the Discord / model-switcher / Use-everywhere topbar chips to icon-only with hover tooltips; the GitHub Star chip stays a full pill. * test(e2e): expand collapsed nav rail before driving rail destinations The entry nav rail now starts collapsed, so the UI smoke tests that click `entry-nav-*` items (or assert the rail/logo is visible) timed out. Add an idempotent `ensureRailOpen` helper and call it before rail nav interactions across the e2e/ui suite; adjust the two rail-structure assertions to expand first. * test(e2e): reach home via Home nav and expand rail for help launcher The logo now doubles as a hover-to-collapse control, so clicking it to return home is intercepted by the collapse overlay — route via the explicit Home nav item instead. Also expand the rail before opening the rail-footer help menu, and guard the execution-pill nav loop with ensureRailOpen. * fix(web): make collapsed rail inert and restore Discord chip title Address review feedback: - The collapsed nav rail kept its logo + nav buttons mounted and focusable, so a fresh Tab on the home view landed on invisible rail controls before the visible toggle/hero. Mark the rail `inert` while collapsed so its descendants leave the keyboard tab order and pointer flow; add a keyboard regression test. - Restore the Discord chip's native `title` alongside `data-tooltip` so the existing topbar contract test (entry-topbar.test.ts) keeps passing. * fix(web): make collapsed templates gallery inert and keep rail foldable on touch Address second review pass: - The collapsed new-user templates gallery was only aria-hidden + height-clipped, so its still-mounted buttons/tabs/search stayed in the keyboard tab order. Mark the gallery body `inert` until revealed (deps include `enabled` so it applies once the async-mounted body appears); add a keyboard regression test. - On coarse-pointer / non-hover devices the hover-only collapse button plus the display:none topbar toggle could strand the rail open. Keep the collapse control visible + tappable under `@media (hover: none)`; add a CDP-emulated touch regression test. * fix(web): keep inline switcher chip accessible name in icon-only topbar --------- Co-authored-by: qiongyu1999 <2694684348@qq.com> |
||
|
lefarcenandGitHub
|
aba4b5bff2 |
fix(home,daemon): keep mode card on-screen and let plugin authoring pause on a question-form (#3640)
* fix(home,daemon): keep mode card on-screen and let plugin authoring pause on a question-form Two release/v0.10.0 firefight fixes: - Home Design Agent description card: the mode toggle's popover card is only shown on the home hero and anchors upward, so at its default 360px it crowds (and on smaller windows clips) the page header. Cap its height on the home hero so it always clears the top chrome and scrolls for the overflow. - Plugin authoring 'ended before generating the required generated-plugin artifacts' error: the od-plugin-authoring turn-1 flow emits a <question-form> to collect the brief, then stops and waits for the user. That turn legitimately closes with code 0 and no artifacts yet, but the missing-artifacts guard treated it as a failure. Track whether the agent emitted a clarifying <question-form> in its streamed text and skip the guard in that case. Red spec: chat-route.test.ts 'does not fail plugin authoring when the turn-1 reply is a clarifying question-form awaiting the brief' (red on main, green on branch). * fix(daemon): recognize <ask-question> alias in plugin-authoring clarification guard The web form parser (apps/web/src/artifacts/question-form.ts) accepts both <question-form> and its <ask-question> alias as a valid brief form. The plugin-authoring missing-artifacts guard only matched the canonical tag, so a turn-1 reply using the alias still rendered a valid form in the UI but fell through to the 'ended before generating…' failure. Mirror the parser's accepted tag set in the daemon guard and add a regression test for the alias path. Addresses Looper review feedback on #3640. * fix(daemon): gate plugin-authoring clarification carve-out on a renderable form body The previous carve-out matched only the question-form/ask-question opening tag, so a malformed (non-JSON) body — which the web parser keeps as raw prose with no rendered form card — would still suppress the missing-artifacts failure, turning a hard failure into a false success. Mirror the parser's minimal renderable contract in a daemon-local helper (emittedRenderableQuestionForm): a closed question-form/ask-question block whose body parses as JSON to an object with a non-empty questions array. Buffer the run's visible text and validate at close. Documented as a deliberate mirror of apps/web/src/artifacts/question-form.ts since the app boundary forbids importing it; promote to packages/contracts if they drift. Tests: question-form and ask-question fixtures now use valid JSON bodies, plus a new regression pinning that a non-renderable (non-JSON) form body still fails. Addresses Looper review feedback on #3640. * fix(daemon): keep question-form close-tag scan Unicode-safe in the daemon mirror The mirrored renderable-form check lowercased the whole buffer before indexing the close tag. Some code points expand under toLowerCase ('İ' -> 'i̇'), shifting every offset after them, so closeIdx desynced from openEnd (original-string coordinate) and the JSON body slice was corrupted — failing a valid form when a casing-expanding char preceded it. Mirror the web parser's findCloseTag: scan char-by-char and lowercase each fixed-length candidate slice, returning an index in the original-string space. Added a daemon regression with an 'İ' preamble before a valid form. Addresses Looper review feedback on #3640. |
||
|
|
d6fe99bcdf |
Make existing project folders easier to open and iterate (#3529)
* Support opening and iterating existing project folders * Fix existing folder flow i18n strings * Stabilize AMR status sync --------- Co-authored-by: Ben Stadler <ben@thebenstadler.com> Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
lefarcenandGitHub
|
5555b2eca9 |
fix(web): keep the streaming question-form id stable (#3582 follow-up) (#3644)
* fix(web): keep the streaming question-form id stable (no churn) `parsePartialQuestionForm` derived the form `id` from the streaming body `id` token, but that arrives char-by-char and `parsePartialJson` repairs the open string, so `form.id` churned (`"d"` → `"di"` → …). `id` keys the live, still-editable Questions panel (ProjectView's questionFormKey, QuestionForm's data-form-id / input names), so each partial-id change remounted the panel and discarded in-progress answers. Derive the preview id from the open-tag attr (complete the instant the tag streams) or the stable default instead; the final parse (tryParseForm, after the close tag) still adopts the real body id. This also keeps the earlier nested-id-masquerade guarantee (no body scan). Follow-up to #3582. Regression: the form id stays `discovery` while the body id token streams in. * fix(web): adopt the question-form body id once complete (no preview→final remount) Building on the char-churn fix: when a form omits the open-tag id attribute but carries a body id, the preview used 'discovery' and the final parse adopted the body id — a single remount at the close tag that could drop answers. Adopt the body id as soon as its string literal is terminated (depth-1, nested ids ignored); that value equals what tryParseForm assigns, so preview id == final id and the panel never remounts on the swap. The id settles early (it's a leading field), before any question is answerable. Regressions: a partial body id is not adopted (no churn); a complete body id is adopted and matches splitOnQuestionForms' final id; nested question ids are ignored. * fix(web): decouple the question-form React key from the parsed form id Root-cause fix for the identity churn three reviewers flagged. The Questions panel was keyed on `${conversation}:${message}:${form.id}`, so any change to the parsed id remounted it and dropped in-progress answers. The streaming preview shows the `discovery` fallback id until the body id arrives, and a form that emits answerable questions before its id settles flips identity while the user is mid-answer — exactly the remount they described. There is at most one (first) form per assistant message, so the id is redundant for uniqueness: `${conversation}:${message}` already identifies the occurrence, and a distinct later form lives in its own message (its own key, still replays the reveal). Extract buildQuestionFormKey (id-free) and unit-test that the key is stable across a streaming id flip, distinct across messages, and null until conversation/message/form are all present. completeTopLevelString stays — it settles the *displayed* id early — but the id no longer drives identity, so even the questions-before-id ordering can't remount the panel. |
||
|
lefarcenandGitHub
|
d0bca92646 |
fix(conversation-forking): fork from the in-memory snapshot so an unpersisted turn no longer silently fails (#3648)
"Fork from here" creates a new conversation by asking the daemon to copy the source conversation's messages up to the fork point, looked up in SQLite by `forkAfterMessageId`. When the fork point is an assistant turn whose run errored or had its connection reset before the message was written to the database (common with AMR/ACP "Connection reset by server"), that id is not in the DB. The daemon returned 404 "fork message not found", `createConversation` turned the non-OK response into `null`, and `handleForkFromMessage` threw into a quiet error state — so no new conversation opened and the user never switched into the fork. The page just flashed and stayed put. Make fork copy what the user is actually looking at: the chat "Fork" action now sends the in-memory messages slice (up to and including the fork point) as `seedMessages`, and the daemon prefers that snapshot over reading the source conversation from the DB by id. The Side Chat path (no `seedMessages`) is unchanged and still seeds from `seedFromConversationId`. Red spec: apps/daemon/tests/projects-routes.test.ts "forks from a client-supplied snapshot when the fork point was never persisted" — 404 on main, 200 on this branch. |
||
|
|
3a21392c46 |
Fix Electron composer lag and stale generation failures (#3593)
* Keep chat typing and web CSS resolution responsive Cache the normalized mention index by entity array and scan only concrete @ positions so the composer does not rescan every file/plugin/skill token on each draft update. Link the shared components package at the root importer so Turbopack/PostCSS can resolve the exported component stylesheet from the web CSS entrypoint. Constraint: Chat composer typing updates run on every keystroke in Electron, and tools-dev/Next evaluates global CSS from the workspace root. Rejected: Debouncing textarea state updates | would make text entry feel stale and would not remove the expensive synchronous parser path. Rejected: Relative-importing package source CSS from apps/web | would bypass the package export contract and hardcode package layout into the app stylesheet. Confidence: high Scope-risk: narrow Directive: Keep inline mention parsing proportional to draft mention markers, and keep package CSS imports resolvable through workspace package links. Tested: pnpm install Tested: pnpm --filter @open-design/web test -- tests/utils/inlineMentions.test.ts Tested: pnpm --filter @open-design/web typecheck Tested: pnpm --filter @open-design/web build Tested: pnpm guard Not-tested: Full web Vitest remains blocked by unrelated CSS selector tests observed before this commit. * Keep composer typing off the render path Move keystroke-time draft updates into the textarea/ref path while deferring React draft rendering, mention overlay work, picker filtering, and localStorage persistence so Electron input echo stays within frame budget. Constraint: Electron chat input was visibly lagging for plain text and mention text while the right iframe stayed responsive. Rejected: Keep the textarea fully controlled | every keystroke still synchronizes the largest ChatComposer render path. Confidence: high Scope-risk: moderate Directive: Keep prompt submission, slash commands, and mention insertion reading draftRef when draft rendering is deferred. Tested: pnpm guard; pnpm --filter @open-design/web typecheck; pnpm --filter @open-design/web build; pnpm exec vitest run -c vitest.config.ts tests/components/ChatComposer.infinite-render.test.tsx; pnpm exec vitest run -c vitest.config.ts tests/components/ChatComposer.context-pickers.test.tsx; pnpm exec vitest run -c vitest.config.ts tests/utils/inlineMentions.test.ts; Electron inspect eval plain text avg 0.03ms max 0.5ms; Electron inspect eval @ path avg 0.03ms max 0.1ms; HTTP smoke status 200 with no components CSS resolve error. Not-tested: Full web Vitest suite and Playwright UI suite. * Keep Electron chat input responsive under heavy projects Constraint: Electron project views can carry large chat/file DOM and Chromium can throttle or defer visible renderer work under tab/window visibility edge cases. Rejected: Re-rendering the composer draft on every plain keystroke | it couples textarea input to mention overlay, resize, storage, and project layout work. Confidence: medium Scope-risk: moderate Directive: Keep plain text input on the textarea/ref path; only commit React draft state immediately for mention, slash, chip, or imperative draft mutations. Tested: pnpm guard; pnpm --filter @open-design/web typecheck; pnpm exec vitest run -c vitest.config.ts tests/components/ChatComposer.infinite-render.test.tsx; pnpm --filter @open-design/desktop test; pnpm --filter @open-design/desktop typecheck; desktop inspect actual project composer with OS SendKeys. Not-tested: Push/PR remains blocked by GitHub 403 for mimicryluden on nexu-io/open-design. * Keep heavy project typing isolated from preview layout Constraint: Electron project panes can carry large DOM histories and preview frames that make textarea paint sensitive to parent layout. Rejected: Further debouncing ChatComposer input state | app-side handler costs were already low while native textarea updates still lagged. Confidence: high Scope-risk: moderate Directive: Keep the composer mounted through the fixed layer/slot pair so future layout changes do not pull typing back into the split-pane flex reflow path. Tested: pnpm --filter @open-design/web typecheck; pnpm exec vitest run -c vitest.config.ts tests/components/ChatComposer.infinite-render.test.tsx tests/components/ChatComposer.context-pickers.test.tsx; pnpm guard; pnpm --filter @open-design/web build; desktop inspect fixed-layer DOM and SendKeys probe avg beforeinput->input 1.4ms max 2.5ms Not-tested: PR publish blocked by GitHub repository permissions if remote still rejects mimicryluden * Keep stale run recovery from hiding generated previews Constraint: Completed daemon runs can outlive the in-memory run registry after a restart. Rejected: Replaying succeeded rows with empty producedFiles | It turns completed messages into failed messages when the registry is gone. Confidence: high Scope-risk: narrow Directive: Only active run statuses should enter daemon reattach recovery. Tested: vitest generation-preview; vitest ProjectView reattach restore; web typecheck; pnpm guard; web build; desktop DOM eval failureText=false iframeCount=2 Not-tested: Fresh provider generation against an external model * Keep composer contracts aligned with latest main Rebase conflict resolution restored the current composer contract while leaving the lower-risk latency fixes in separate layers. Constraint: The branch was rebased over newer workspace/session context composer contracts. Rejected: Keeping the older composer conflict side because it removed current ChatComposer props and broke web typecheck. Confidence: high Scope-risk: narrow Directive: Preserve upstream ChatComposer contracts when rebasing input-latency work. Tested: web typecheck; targeted ChatComposer, inlineMentions, ProjectView generation-preview tests; desktop window-chrome test Not-tested: Full browser interaction after fork push * Fix daemon Nix workspace symlink pruning --------- Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
lefarcenandGitHub
|
4e17440f81 |
fix(web): keep empty persisted folders visible in Design Files (#3358 regression) (#3631)
* fix(web): keep empty persisted folders visible in Design Files
`folders` prop, so an empty persisted folder (user-created, or an empty dir in
an imported working directory) vanished — it couldn't be opened or deleted
until a file lived inside it. Thread the fetched `folders` back into
DesignFilesPanel and merge them into `dirsAtCurrentDir`; don't show the
project-empty state when folders exist; and treat a persisted folder as a
valid currentDir so navigating into an empty one doesn't bounce to the root.
FileWorkspace passes `folders={projectFolders}` (already fetched).
Regressions: an empty persisted folder shows at the root, and a nested empty
folder is navigable.
(cherry picked from commit 8f45bb14ccd14b04fb281c3b0eea06a0f5021446)
* fix(web): reset project folders when switching projects
With persisted folders now rendered, FileWorkspace kept the previous project's
folder list until the async fetchProjectFolders resolved, so switching to a
different (empty) project briefly showed the old folders and suppressed the
new project's empty state. Clear projectFolders synchronously on projectId
change, before the fetch.
Regression in FileWorkspace.test.tsx: rerendering with a new projectId while
the new folder fetch is pending drops the previous project's folders.
* fix(web): reset project folders during render, not in an effect
mrcfps flagged that the effect-based folder reset still let the new
DesignFilesPanel (keyed by projectId) mount once with the previous project's
folders before the effect cleared them — briefly suppressing the new project's
empty state, the exact regression this follow-up removes. Move the reset into
the render phase (adjust state during render via a projectId ref, mirroring the
designFilesNav reset), so the new panel never observes stale folders. A stable
EMPTY_PROJECT_FOLDERS constant keeps the render-phase setState idempotent.
Tightens the regression: a wrapper records the folders prop on every
DesignFilesPanel render, and the test asserts no render for the new project
ever sees a non-empty folder list — the pre-paint frame RTL's post-rerender DOM
check can't observe. Red on the effect-based reset, green here.
|
||
|
Tom HuangandGitHub
|
fd1c0ecb93 |
Feat/optimize style (#3624)
* Refactor ChatComposer and DesignBrowserPanel components for improved functionality and localization - Introduced new functions in ChatComposer for managing attachment orders and appending staged attachments, enhancing user experience during file uploads. - Updated ChatPane to clarify the search button's aria-label for better accessibility. - Enhanced DesignBrowserPanel with new localization keys for browser use inputs and actions, improving clarity across multiple languages. - Adjusted styles in various components for better layout and user interaction, including updates to the chat input and staged context elements. * Update AvatarMenu and Chat components for improved functionality and localization - Modified AvatarMenu to include an optional section parameter for the onOpenSettings callback, enhancing flexibility in opening settings. - Updated tests for AvatarMenu to verify the correct section is passed when opening settings. - Adjusted styles in various components, including the ChatComposer and FileViewer, to improve layout and user interaction. - Enhanced localization strings across multiple languages to reflect changes in chat mode terminology from 'Chat' to 'Ask', improving clarity for users. |
||
|
goododeo-specandGitHub
|
18169ee52a |
Improve export fidelity contract for editable PPTX and PDF readiness (#3628)
* Improve export fidelity contract * Improve export fidelity contract * Improve export fidelity contract * Improve export fidelity contract * Wait for CSS images in desktop print handshake * Wait for CSS images in desktop print handshake |
||
|
lefarcenandGitHub
|
6e10fa4852 |
feat: stream agent questions token-by-token (#3582)
* feat(web): stream AskUserQuestion and discovery questions token-by-token Render both agent question surfaces as they generate instead of popping in only once the whole payload finishes. - AskUserQuestion (Claude tool): builds on the live tool-input streaming landed in #3382. When a still-streaming tool's name is AskUserQuestion, render a read-only StreamingAskUserQuestionCard that grows its questions and options from the repaired JSON prefix; the interactive (persisted) card takes over once the full tool_use lands and the id leaves liveToolInput. - <question-form> discovery (any agent, incl. Codex): the streaming parser now repairs the truncated prefix so a question shows the moment its label exists and its options grow in one at a time, instead of waiting for each question object to fully close. The Questions panel UX (skip countdown, auto-continue, reveal) is unchanged. Shared repairJsonPrefix/parsePartialJson live in runtime/partial-json.ts. No daemon, contract, or provider changes — this reuses #3382's tool_input_delta / liveToolInput plumbing. * fix(web): harden streaming <question-form> parser (review follow-ups) Two merge-safe follow-ups from review on parsePartialQuestionForm: - Derive form-level id/title/description from the parsed top-level object instead of regex-scanning the whole partial body, so a nested question or option id/description can no longer masquerade as the form's own. form.id keys the live Questions panel (ProjectView), so a mid-stream identity change would remount it. Removes extractJsonStringField. - Strip a (possibly partial) trailing ``` fence so the live preview keeps streaming in the gap between the fenced JSON body closing and the </question-form> close tag arriving, instead of dropping back to empty. Both covered by regression tests in question-form.test.ts. * fix(web): preserve submitLabel + string-safe fence trim in streaming form parser Second round of review follow-ups on parsePartialQuestionForm: - Carry top-level submitLabel through the streaming preview, mirroring tryParseForm, so a custom CTA label doesn't flicker in only once the close tag arrives. - Only strip the trailing ``` fence when those backticks are the closing wrapper, not content of a JSON string value still being typed — a label like "Use ```" mid-stream kept losing its backticks. The trim now checks it isn't inside an open JSON string first. Both covered by regression tests in question-form.test.ts. * fix(web): honest status badge on WebFetch/WebSearch cards WebFetchCard / WebSearchCard hard-coded a green check badge regardless of the real tool state, so a slow or failed fetch/search rendered as false-green. Thread result / runStreaming / runSucceeded through both and reuse ResultBadge (as the other tool cards do) so the badge reflects the actual outcome. Pre-existing on main (only surfaced in this PR's changed range because the new streaming card additions shifted these lines down); folded in here since it's a one-line-per-card fix in a file this PR already touches. * fix(web): keep partial-JSON repair stable across a dangling string escape repairJsonPrefix() appended a closing quote even when the stream stopped on an incomplete escape inside an open string, leaving invalid JSON: `{"a":"x\` became `{"a":"x\"` (the quote got escaped) and `{"a":"x\u` became an invalid `\uXXXX`. parsePartialJson() then returned null, collapsing the live AskUserQuestion / question-form preview to its empty frame mid-stream whenever a prompt mentioned a Windows path, regex, or escaped quote. Now neutralize a trailing incomplete escape before closing the string: drop a partial `\uXXXX` (0-3 hex) and a lone (odd-run) trailing backslash. Completed escapes and even backslash pairs (e.g. `C:\\Users\\`) are preserved. Regression tests in ask-user-question.test.ts cover the repair cases and a path-like prompt that keeps its last visible content. * fix(web): trim unfinished scalar values in partial-JSON repair repairJsonPrefix() only trimmed dangling commas / empty `"key":`, so a delta that stopped mid-scalar left invalid JSON: `{"…","multiSelect":f` closed to `{…:f}` and failed to parse, collapsing the live preview whenever a boolean, null, or number token split across deltas. The cleanup loop now also trims a partial `true`/`false`/`null` and a number cut on `.`/`e`/sign, leaving the `"key":` / `,` / `[` boundary for the next iteration. Complete literals (`true`, `12`, `1.5`, `1e3`) and string values that merely start like a literal are not prefixes of these patterns, so they survive. Regressions added in both the AUQ and question-form suites (a boolean split across deltas keeps already-visible questions) plus repair-level unit tests. * fix(web): stable preview question ids while a form streams shapeStreamingQuestions surfaced a question as soon as its label existed, but mapRawQuestion falls back to q${index+1} until the real id streams — so a question that arrives label-first and gains its canonical id later would change id mid-stream. The Questions panel stays editable while building and keys fields / answers off q.id, so that id change remounts the field and orphans whatever the user already picked. Pin the preview id to the array index for the whole streaming phase (questions only append once their label exists, so the index is stable); the final non-preview parse still uses the real id. Regression in question-form.test.ts: label-first then id-later keeps id q1. * fix(web): preview question ids match the final parse (no orphaned answers) The previous index-pin kept ids stable *within* the preview but still differed from the final canonical id, so an answer entered while the panel was building got stranded at the preview→final swap (field re-keys from `q1` to `platform`, reads nothing). Surface a streaming question only once its canonical id is determinable instead: use the streamed `id`; or, for an already-closed (not in-flight) object with no id, the `q${index+1}` fallback that the final parse will also assign; hold the in-flight object until it gains an id. Preview id now always equals the final id. Updated/added regressions in question-form.test.ts: a label-first object is held until its id is determinable, and preview ids match splitOnQuestionForms. * fix(web): stable positional keys in the streaming AskUserQuestion card The streaming card keyed fields on q.question and options on opt.label, but those strings grow token-by-token, so every delta changed the key and remounted the subtree — replaying the reveal animation on each token instead of mounting once. Switch to positional keys (questionIndex / questionIndex:optionIndex); questions and options only append during streaming, so the index is stable and each node updates its text in place. Regression: StreamingAskUserQuestionCard.test.tsx asserts the field/option DOM nodes are reused (===) across successive grown prefixes. * fix(web): stream a closed no-id question before the close tag The previous gate held *every* last no-id question as if in-flight, but a last object whose braces have already streamed has a final fallback id (q${index+1}) and should appear immediately — holding it broke the "question shows the moment its label exists" behavior for forms that omit explicit ids. Distinguish closed vs still-open via a string-aware balanced-object count: surface a question once its object has closed (id final) or it already carries an `id` (stable); keep holding only an in-flight object with no id yet. Regressions: a single closed no-id question streams pre-close-tag (id q1), while the same content with an unclosed object stays held. * fix(web): suppress hedging fallback text while a live AskUserQuestion streams suppressAskUserQuestionFallbackText ran on the persisted blocks before the liveBlocks pass, so during streaming (no persisted tool_use yet) seenAskUser Question never flipped and any duplicate hedging markdown survived into persistedBlocks while the live card was appended after — the exact duplicate-question state the card removes, just earlier in the turn. Seed the suppressor with whether a live AskUserQuestion is present in liveToolInput so it fires during the streaming phase too. Regression in AssistantMessage.test.tsx: fallback prose coexisting with a live AskUserQuestion in liveToolInput is suppressed, while normal prose without a live AUQ is kept. * fix(web): order-aware fallback suppression for live AskUserQuestion The previous fix seeded the suppressor with "a live AskUserQuestion exists", which dropped ALL text blocks in the turn — including legitimate preamble ("I need one clarification:") that precedes the tool call, not just duplicate hedging. Make it order-aware instead: suppress text only after an AskUserQuestion in block order, and have the suppressor recognize a live (streaming) AskUserQuestion live-tool block the same way it recognizes a settled tool-group. Preamble before the card is kept; hedging after it is dropped — matching the settled-case semantics. Regression updated in AssistantMessage.test.tsx: intro preamble coexisting with a live AskUserQuestion in liveToolInput is preserved. * fix(web): place the live AskUserQuestion at its real stream position The order-aware pass still appended the live AUQ block after all persisted blocks, so hedging text the model emits *after* starting the tool call (which is already in events, hence before the appended block) was kept. Carry the event index the tool started at (`seq`, recorded in liveToolInput when the first input_json_delta arrives) and split the events around it: blocks before `seq` are preamble, the live AUQ block goes next, blocks after `seq` are hedging. Suppression then keeps the preamble and drops the trailing hedging, exactly matching the settled tool_use case. Regression in AssistantMessage.test.tsx: with the tool starting after the preamble (seq=1), preamble before the card is preserved and hedging after it is dropped. * fix(web): account for buffered preamble when pinning live-tool seq `seq` was read from `latestAssistantMsg.events.length`, but preamble text goes through `appendTextEvent` and isn't committed to `events` until the next rAF/250ms flush. When intro prose and the first `tool_input_delta` arrive in the same burst, that read undercounts by one, so the live AskUserQuestion was positioned before the preamble and suppression dropped it as hedging. The buffer now exposes `hasPendingText()`, and seq adds 1 for a still-pending preamble chunk (it commits as one text event just before the tool position). Regression in buffered-text-pending.test.tsx: appended text is reported pending and absent from events until flush, then commits as one event. * fix(web): re-render the streaming row on live tool-input deltas AssistantMessage is memoized with a prop allowlist that omitted liveToolInput, so a burst of pure tool_input_delta traffic (which only changes liveToolInput, not message) was swallowed by the comparator — the live AskUserQuestion card / code box wouldn't grow until some unrelated change, usually the final tool_use. Add liveToolInput to ASSISTANT_MESSAGE_COMPARED_PROPS, and have ChatPane pass it only to the streaming row (undefined elsewhere) so just that row re-renders per delta, not all N messages. Also widens ChatPane's liveToolInput type with the optional seq field. Regression in AssistantMessage.test.tsx: with the same message object, changing only liveToolInput re-renders and grows the card prompt. |
||
|
chenxueandGitHub
|
b1fd529eea |
Feat/aihubmix byok provider (#3583)
* feat(media): integrate AIHubMix as full-stack BYOK provider Add AIHubMix (https://aihubmix.com) as a BYOK provider at the same layer as SenseAudio — both a chat protocol and a media-generation provider — so users can paste one key and get chat + image + speech, with the same key auto-mirrored into media-config for the `od media` CLI path. AIHubMix is an OpenAI-wire-compatible aggregator. Its distinguishing trait is the fixed APP-Code attribution header (discount), funnelled through a single aihubmixHeaders() helper so every outbound call point (chat proxy, connection test, model discovery, media renderers, memory extractor) injects it in one place. Chat protocol layer: - Extend ApiProtocol / ConnectionTestProtocol / MemoryExtractionProvider unions. - New web provider client + dispatch; mirror across apiProtocols, config, utils, Settings/ChatComposer/HomeHero UI surfaces. - Refactor the SenseAudio tool-loop chat proxy into a shared registerByokToolChatProxy() factory; register senseaudio + aihubmix through it (senseaudio behaviour byte-preserved). connectionTest + providerModels treat aihubmix as OpenAI-shaped with APP-Code headers. Media layer: - Register provider + image (aihubmix-gpt-image-1 / -dall-e-3) and speech (aihubmix-tts-1) models; prefixed ids stay unique and map to wire names. - OpenAI-compatible renderers (/v1/images/generations, /v1/audio/speech) and chat tool executors; ENV_KEYS + media-picker allowlists updated. Note: AIHUBMIX_APP_CODE in apps/daemon/src/aihubmix.ts is an empty placeholder — fill it with the code from https://aihubmix.com/appstore to enable the discount header (the integration works without it). Tests: media-aihubmix.test.ts (wire-name mapping, APP-Code invariant, image + speech renderers) and an AIHubMix proxy-route case; verify:media-models, guard, and typecheck all green. * feat(media): source AIHubMix model lists from live catalog endpoints Chat LLM models and media image models now come from AIHubMix's live catalogue (GET /api/v1/models?type=llm | image_generation, public, returns model_id / model_name) instead of only the hand-curated static lists. Chat: the existing /api/provider/models discovery, when protocol=aihubmix, now hits the type=llm catalogue and parses the AIHubMix envelope (342 models live). The static SUGGESTED_MODELS remain as the pre-fetch default. Media image: new GET /api/media/providers/aihubmix/models endpoint proxies the type=image_generation catalogue (53 models), prefixes ids `aihubmix-` for uniqueness, and caches for 5 min with stale-on-error fallback. generateMedia synthesizes a def for any unregistered `aihubmix-*` id (like the fal-ai/* path) so every catalogue model renders through the OpenAI-compatible image endpoint. Helpers (aihubmixCatalogUrl / parseAIHubMixCatalog) + tests for url building, envelope parsing, and rendering a dynamic catalogue id. The web media pickers still read the static seed list; wiring them to the live endpoint is a follow-up UI change. * feat(web): show live AIHubMix image catalog in all media pickers Wire the AIHubMix image-generation catalogue (53 models, fetched from the new daemon endpoint) into every image-model picker. A shared hook (useAIHubMixImageModels) fetches once per session and caches; mergeAihubmixImageModels replaces the two static aihubmix seeds with the live list (and falls back to the seeds when the fetch hasn't resolved or fails, so the picker is never empty). Surfaces: - NewProjectPanel image picker — merged list passed to supportedModels('image'). - Home media composer — buildHomeMediaComposer / fieldsForSurface / normalizeHomeMediaInputs now accept an injected imageModels list; HomeView fetches and threads it through all three composer call sites. Saved/template image-model ids starting with `aihubmix-` are accepted even though they're absent from the static IMAGE_MODELS, since they render dynamically daemon-side. * feat(byok): load AIHubMix chat model list without an API key The BYOK chat model dropdown already merges live-fetched models with the static suggestions, but the fetch was gated on a non-empty API key (every provider hits /v1/models behind its key). AIHubMix's catalogue is public, so its 342-model list now loads as soon as the AIHubMix tab is selected — no key required. - Daemon /api/provider/models accepts an empty apiKey when protocol=aihubmix (baseUrl still required); providerModelsHeaders sends no Bearer when the key is blank. - Web: missingByokModelFetchFields treats apiKey as optional for aihubmix, and the auto-fetch effect skips the commit-on-blur gate for aihubmix (base URL is prefilled and there's no key to debounce), so the list populates on tab select. * feat(byok): default AIHubMix chat model to claude-opus-4-8 * feat(byok): memory model picker uses live searchable model list The Memory model picker was a plain <select> hard-wired to the 8 static SUGGESTED_MODELS. Swap it to the same SearchableModelSelect the chat picker above uses, fed by the parent's merged fetched+suggested model options (apiModelOptions). For AIHubMix that means the full live 342-model catalogue shows as a searchable dropdown (keyless), matching the chat picker; other protocols get their live /v1/models list once the chat key is set. 'Same as chat' leads the list and 'Custom…' is pinned last. * feat(byok): BYOK chat image-model picker lists live AIHubMix catalog The Settings BYOK 'image generation model' dropdown (byokImageModel, used by the chat-injected generate_image tool) listed only the 2 static aihubmix seeds. Feed it the live image catalogue (useAIHubMixImageModels) for aihubmix so it shows all ~53 models; SenseAudio keeps its static registry list. Relax the daemon's isAIHubMixImageModel to accept any aihubmix-* id (they all render via the OpenAI-compatible endpoint with prefix-strip), so a dynamically-chosen default passes validation instead of being dropped back to the seed default. * refactor(byok): share one hook for BYOK image-model pickers; wire composer The chat composer's inline 'image generation model' picker still listed only the static aihubmix seeds. Rather than re-derive the merge+filter a third time, extract useByokImageModelOptions(provider) into the shared aihubmix-image-models module (live AIHubMix catalogue for aihubmix, static registry otherwise) and use it from both the Settings image picker and the composer picker. One cached fetch feeds all sites. * feat(byok): route AIHubMix chat by model name to native protocol wire Per the AIHubMix integration guide §4.3, route the BYOK chat by model name to each family's native endpoint on the AIHubMix origin instead of forcing every model through /v1/chat/completions: claude* → Anthropic /v1/messages (x-api-key + anthropic-version) gemini*/imagen* → Gemini :streamGenerateContent (x-goog-api-key) everything else → OpenAI /v1/chat/completions (keeps the tool loop) All three carry the APP-Code attribution header. The claude/gemini routes run base chat only (no in-chat generate_image yet) — each protocol has a pluggable tool slot for a later Anthropic tool_use / Gemini functionCall adapter. The OpenAI family keeps in-chat image/speech generation. Reuse: extracted runAnthropicChatStream / runGeminiChatStream (+ payload builders) from the existing /api/proxy/{anthropic,google}/stream handlers; those routes now delegate to the same functions (behaviour preserved, 79 proxy tests green). New helpers classifyAIHubMixModel / aihubmixOriginFromBase / aihubmixAppCodeHeader in aihubmix.ts. No SDK added. Default AIHubMix chat model is now gpt-5.5 (an OpenAI-family model) so the default keeps in-chat generate_image working after routing. Tests: claude→/v1/messages and gemini→/gemini/.../streamGenerateContent routing assertions in proxy-routes.test.ts. * feat(byok): AIHubMix in-chat image/video/audio model selection + media-adapters Bring the BYOK chat media surface to parity across image, video and audio: - Composer + Settings dropdowns for image/video/speech models and TTS voice, threaded as byokVideoModel / byokSpeechModel / byokSpeechVoice. The user's explicit pick is authoritative over the LLM's tool `model` arg (regression the LLM kept overriding the composer choice). "Default" resolves to the first live-catalogue model (honest label) with the daemon const as final fallback. - New apps/daemon/src/media-adapters: injection-style capability registry + family-aware video request builder (seedance content[] array vs generic input_reference), capability shape aligned with OpenRouter /videos/models. The chat video executor now builds requests through it. - Per-family request shaping: * video: duration snapped to each family's allowed set (veo 4/6/8, sora 4/8/12, wan 5/10); i2v reference resolved to a data URL. * image: gemini/nano-banana models routed to gemini-native generateContent (the /v1/images/generations shape 400s for them); others stay OpenAI-shaped. * speech: gemini TTS via generateContent with PCM->WAV wrapping and Gemini voice fallback; others use /v1/audio/speech (mp3). * download: same-origin AIHubMix asset URLs re-send Bearer + APP-Code. - Clear, actionable errors for unsupported i2v / "params dropped" models, plus per-call submit logging (model + endpoint) for multi-model verification. - i18n keys across 19 locales for the new pickers and the neutral default label. Tests: byok-tools + media-adapters specs (85 passing); guard + daemon/web typecheck green. * feat(byok): media tool loops on claude/gemini chat + wan i2v wire + seedance resolution fix Make AIHubMix in-chat media generation work regardless of which chat model is selected, and fix two image-to-video wire-shape bugs surfaced by live testing. chat-routes: the aihubmix routeByModel divert sent claude -> Anthropic /v1/messages and gemini -> Gemini generateContent BEFORE the OpenAI tool loop, so those chat models had no generate_image/video/speech tools ("没有音频工具"). Add per-protocol tool loops inside registerByokToolChatProxy (reusing the shared executeOneTool): runAnthropicToolTurn/Chat parse tool_use content blocks + input_json_delta; runGeminiToolTurn/Chat parse functionCall parts. The shared single-pass runAnthropicChatStream/runGeminiChatStream and the direct /api/proxy/{anthropic,google}/stream routes are untouched. Factor the per-kind tool-result hint into buildToolResultContent (shared by all three protocols) and add openaiToolsToAnthropic / openaiToolsToGemini schema converters (Gemini schema sanitized to the OpenAPI subset). Gemini 3.x attaches a thoughtSignature to each functionCall part that must be echoed back verbatim or the follow-up 400s, so replay the raw functionCall part instead of reconstructing {name,args}. media-adapters: seedance accepts resolution only as a token (480p/720p/1080p), never a WxH pixel string — passing the aspect-derived 1280x720 made it 400 ("resolution ... not valid ... in i2v"). Add snapResolutionToken to normalise. Add a third request family `wan` for wan*/happyhorse*: verified against a working happyhorse-1.0-i2v call, these are DashScope/wanx-backed and use { input:{ prompt, media:[{type:first_frame,url}] }, parameters:{ resolution, duration, prompt_extend, watermark } }, NOT the flat input_reference shape we were sending. Update the i2v failure-advice message accordingly (the remaining risk is wanx needing a public image URL while we send an inline data URL; seedance accepts the inline image). Tests: anthropic+gemini in-chat tool loops (incl. thoughtSignature round-trip), snapResolutionToken, seedance token resolution, and the wan input.media shape. * feat(byok): add veo video family + default to seedance-2.0-fast veo* models go through the gateway's OpenAI→Gemini predictLongRunning shim, which wants `seconds` as a NUMBER and only honours `size` (no aspect_ratio) — distinct enough from the generic flat shape to warrant its own `veo` family in the media-adapters builder. Verified against a working veo-3.1-lite call. Also default the BYOK chat video model to doubao-seedance-2-0-fast-260128 (the broadly-available fast i2v/t2v path) and extend the media-adapters tests to cover the veo family routing + duration snapping (4/6/8). * feat(byok): hide account-source model labels for AIHubMix AIHubMix's live model fetch returns its full account catalogue, so every picker option carried an identical "from your account" badge that added noise without distinguishing anything. Introduce a reusable ACCOUNT_MODEL_SOURCE_LABEL_HIDDEN set + hidesAccountModelSourceLabel() predicate: for listed protocols the per-option source label is dropped and the loaded-models hint shows a plain count (new settings.modelsLoadedCount key) instead of the account wording. Other providers are unaffected; add a protocol to the set to opt it in. * fix(byok): close AIHubMix catalogue SSRF + share gemini image branch Two review fixes on the AIHubMix BYOK provider: - SSRF: the public `/api/media/providers/aihubmix/models` GET took a caller-controlled `baseUrl` and fetched it server-side, so `?baseUrl=http://169.254.169.254/` could reach a private/metadata host. The catalogue lives at one fixed origin, so the route now ignores any `baseUrl` query and hard-codes the official AIHubMix origin — no attacker-chooseable fetch target remains. A focused route test asserts a malicious `baseUrl` is ignored and only aihubmix.com is contacted. - Gemini image parity: the media renderer always sent AIHubMix image models to the OpenAI `/images/generations` endpoint, which 400s for gemini/imagen-family models. Extract the chat tool's Gemini-native branch into a shared `aihubmixGeminiImageBytes` helper and route those models through it from both the chat tool and the Home/New Project/CLI media path, so the catalogue the picker exposes is fully renderable. Added a media-path test that asserts a gemini-family id hits the generateContent wire. * feat(byok): hide redundant gateway preset for single-origin AIHubMix The "gateway preset" dropdown only ever offered AIHubMix itself (plus the generic custom-endpoint option) because AIHubMix is a fixed-origin gateway with exactly one preset — redundant UI. Gate it behind a reusable GATEWAY_PRESET_HIDDEN set + hidesGatewayPreset() predicate (mirroring the existing ACCOUNT_MODEL_SOURCE_LABEL_HIDDEN pattern) so another single-origin gateway can opt in by adding its protocol, rather than hardcoding a check. The Base URL field still shows the resolved origin; multi-preset and custom-endpoint-reliant protocols (azure/google/senseaudio) are unaffected. * feat(byok): hide Base URL field too for fixed-origin AIHubMix gateway AIHubMix resolves its Base URL automatically, so the editable field was as redundant as the gateway preset. Generalize the just-added GATEWAY_PRESET_HIDDEN into one semantic concept — FIXED_ORIGIN_GATEWAYS + isFixedOriginGateway() — and drive BOTH the preset dropdown and the Base URL field off that single predicate. Adding a fixed-origin gateway protocol now hides both surfaces in one place; the user only supplies an API key. azure/google/senseaudio are unaffected. * fix(byok): close AIHubMix asset-download SSRF redirect bypass The AIHubMix asset downloads validated the literal URL via assertExternalAssetUrl but then fetched with the default redirect:'follow', so a validated public URL could 302 the daemon into 169.254.169.254 / RFC1918 space (and leak Bearer/APP-Code to the hop target). Add a shared assertAndFetchExternalAsset(url, init) helper that validates and pins redirect:'error', and route every external asset download through it: chat video/image (byok-tools.ts) and media image/video + SenseAudio image (media.ts). Also add the missing isLocalSameOrigin guard to the AIHubMix catalogue route so it matches its sibling media routes. Regression tests lock the invariant: helper-level (blocks literal metadata host, forces redirect:'error') plus chat video/image executors asserting the download fetch carries redirect:'error', and a cross-origin rejection test on the catalogue route. Also point the AIHubMix API-key console link at the referral URL. * feat(byok): carry New Project media-model pick into BYOK conversation The model chosen in New Project → Media was stored on project.metadata but dropped for BYOK providers (AIHubMix): media is produced by the generate_* chat tools whose default comes from the per-request byok*Model field, which ProjectView only seeded from Settings. Seed those overrides from the project's creation-time pick instead, gated to the active BYOK provider so a cross-provider model (e.g. a SenseAudio run whose image project defaulted to gpt-image-2) can't clobber the user's Settings default. mediaModelProviderId resolves AIHubMix's `aihubmix-` live ids by prefix so the seed survives before the async catalogue loads. * refactor(byok): remove in-chat media model selector pending unified picker Drop the in-conversation BYOK image/video/speech model dropdowns from the chat composer. Media model selection now comes from the project's creation-time pick (seeded into the byok*Model overrides), and a unified selection surface will replace this later. Also removes the now-dead option hooks/imports so the composer no longer fires AIHubMix catalogue fetches it doesn't render. The byok* props/handlers are retained as plumbing for the future unified picker. * fix(byok): resolve AIHubMix fixed-origin base URL so model lists load Hiding the Base URL field for the fixed-origin AIHubMix gateway left config.baseUrl empty, which blocked the live model-list fetch (it gates on a valid base URL) so the Settings dropdown and the top-bar switcher showed only the static suggested models. Centralise the fixed-origin concept in state/apiProtocols (FIXED_ORIGIN_GATEWAYS / isFixedOriginGateway / resolveFixedOriginBaseUrl) and resolve the canonical origin in loadConfig so every consumer sees it — including configs already persisted with an empty baseUrl. SettingsDialog now imports the shared helpers instead of redefining them. |
||
|
|
0e6d3176ab |
fix(web): prompt injection warning chip + subdirectory files and tab state (#2804)
* fix(web): show subdirectory files in root list and preserve design-files nav state across tabs Two related bugs in DesignFilesPanel: 1. Since directory navigation was added (#2442), files nested in subdirectories were excluded from the root-level flat list, making the page-size select appear broken — changing it could not surface files the user knew existed. Fix: at the root level (currentDir === ''), include all files in filesAtCurrentDir so pagination covers the whole project. Subdirectory rows still render as navigation shortcuts. 2. Switching to any other tab unmounted DesignFilesPanel, resetting kindFilter, currentDir, page, and pageSize back to defaults on return. Fix: FileWorkspace persists the four nav-state fields in a ref and passes them as initial values (navState prop) plus a change callback (onNavStateChange) to DesignFilesPanel. The ref resets when projectId changes, matching the existing key={projectId} remount behaviour. Adds a regression test for case 1. * feat(web): add possiblePromptInjection i18n key to all locales * feat(web): add injection variant CSS for system-reminder chip * test(web): add failing tests for prompt injection chip * test(web): fix null safety and strengthen assertions in injection chip tests * fix(web): render model-echoed system-reminder as prompt injection warning chip Add alert-triangle to Icon, update SystemReminderBlock to accept a variant prop (trusted|injection), render injection variant with amber warning styling and "Possible prompt injection" label, and fix the expand test to wrap click in act() for React 18 batching. * chore(web): document SystemReminderBlock variant semantics * fix(daemon): prepend prompt injection resistance block to every agent system prompt * fix(web): preserve page number across tab switches in DesignFilesPanel Three useEffect hooks that reset page to 0 (on pageSize, kindFilter, and currentDir changes) also fired on initial mount, overwriting the page restored from navState. Added a skipPageResetOnMount ref — true during the first render, set to false by an effect declared after the reset effects — so user-initiated changes still reset the page while navState is honoured on remount. Adds a regression test: navigate to page 2, unmount, remount with the captured navState, assert page 2 is restored. * fix(web): translate possiblePromptInjection key into all 17 non-English locales Replaces the English placeholder 'Possible prompt injection' with localized strings so the warning chip label is readable for non-English users. * fix(web): reset design files nav before project switch render * fix(web): restore prompt injection warning styling --------- Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
|
e194e784c9 |
fix AMR startup error visibility (#3612)
Co-authored-by: Deepak kudi <deepakkudi23@adsl-172-10-9-116.dsl.sndg02.sbcglobal.net> |
||
|
Ray XiandGitHub
|
2e23bafb2b | Implement Langfuse phase 2 telemetry (#3616) | ||
|
|
7403dc681e |
refactor: move daemon routes into routes dir (#3568)
* refactor: move daemon routes into routes dir * fix: restore components typecheck dependencies * fix: update host tools route test import Generated-By: looper 0.9.4 (runner=fixer, agent=codex) * fix: restore review follow-up patches * fix: tighten daemon route typing --------- Co-authored-by: Looper <looper@noreply.github.com> |
||
|
|
e77c739e4e |
feat(daemon): resume Claude CLI sessions across turns (+ skip unchanged instruction block) (#3448)
* feat(daemon): add agent_sessions store for per-conversation CLI session ids * refactor(daemon): align getAgentSession row cast with DbRow convention * feat(daemon): claude adapter emits --resume/--session-id from runtime context * feat(daemon): add resolveAgentResumeContext resume/create decision helper * feat(daemon): detect claude --resume missing-session failures * feat(daemon): resume Claude CLI session across conversation turns * fix(daemon): persist claude session id after spawn-budget guards * feat(daemon): store stable instruction hash on agent_sessions * feat(daemon): carry stored stable-instruction hash in resume context * feat(daemon): skip unchanged stable instructions on Claude resume turns * refactor(daemon): extract computeIncludeStable with unit tests; harden agent_sessions ALTER guard * fix(daemon): persist Claude session after successful turn --------- Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
|
9ac50b34a7 |
fix(web): add BYOK Local CLI recovery (#3391)
Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
|
ccc165f206 |
fix(daemon): don't fail a run when a post-completion SessionEnd hook exits non-zero (#3373)
* fix(daemon): don't fail a run when a post-completion SessionEnd hook exits non-zero A chat run whose model already emitted a clean terminal turn was marked failed when the spawned agent's SessionEnd hook (e.g. the Honcho memory plugin) exited non-zero during teardown, dragging the claude child to exit 1. classifyChatRunCloseStatus mapped any non-zero exit to failed, so a post-result teardown exit surfaced a red AGENT_EXECUTION_FAILED banner and halted the flow for work that had already succeeded. Add a post-completion carve-out: track turnCompletedCleanly at the exact point the daemon observes a clean terminal turn (turn_end/usage with no outstanding host answer -- the same condition that closes the child's stdin), and classify a later non-zero exit or kill signal as succeeded. Non-zero exits before a clean turn still fail, and the agent-specific auth/quota/AMR guards still run first. Fixes #3372 * fix(daemon): record clean turn after host answers --------- Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
lefarcenandGitHub
|
831781141f |
fix(web): create files in the open Design Files folder, not the root (#3358 regression) (#3623)
#3358's panel redesign dropped the `onCurrentDirChange` handoff, so `FileWorkspace.uploadDir` stayed at the project root while `DesignFilesPanel` tracked the viewed folder only in local state. After navigating into a subfolder, Upload / Paste / New sketch / dropped files all created at the project root instead of that folder. Restore the sync: `DesignFilesPanel` reports its `currentDir` via a re-added `onCurrentDirChange` prop, and `FileWorkspace` feeds it back into `uploadDir`. Regression in DesignFilesPanel.test.tsx: navigating into a folder reports the nested dir so the parent creates files there. |
||
|
|
fa80a084ec |
fix: keep saved comment marker order stable (#3615)
Co-authored-by: Deepak kudi <deepakkudi23@adsl-172-10-9-116.dsl.sndg02.sbcglobal.net> |
||
|
|
1a582bf8c2 |
feat(web): add next-step actions after artifact generation (#3439)
* feat(web): add next-step actions after artifact generation Surface a small affordance under the last assistant message once it has produced a previewable HTML artifact, so users have an obvious next move instead of stalling or leaving after the first version. It offers recommended iteration directions (make it more premium, strengthen brand style, tighten copy, generate a second version) as multi-select chips that prefill the composer (not auto-send, so the user reviews before sending), plus a peer "Share" action that reuses the preview workspace's existing share/export menu. There is deliberately no vague "continue editing" / "optimize visuals" catch-all — free-form follow-ups belong in the composer and the visual directions are already covered by the concrete chips. Adds a next_step analytics event (exposure + share/chip clicks with chip_id) to drive the second-turn and share-rate acceptance metrics. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(web): point next-step Share at the deploy/share menu, not download The toolbar's "Share" button opens the deploy menu (share-link + publish online), while the download menu is export-to-disk. Sharing requires a published link, and the publish step lives in the deploy menu, so the next-step Share action should open that surface — not the export one. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(web): preserve next-step chip button semantics * fix(web): preserve browser tabs on artifact share --------- Co-authored-by: chaoxiaoche <chaoxiaoche@chaoxiaochedeMacBook-Pro.local> Co-authored-by: Cursor <cursoragent@cursor.com> |
||
|
|
6408937564 |
feat(web): redesign Design Files panel + pick working directory from Home (#3358)
* feat(web): align Design Files panel with Claude reference UI Redesign the Design Files panel into a continuous grouped table that mirrors Claude's own Design Files surface: uppercase gray section bands (Folders / Stylesheets / Scripts / Documents / Images …), flush full-width rows with a colored icon block + bold name + type subtitle + right-aligned timestamp, a clean single-line toolbar (breadcrumbs left, New sketch / Paste / Upload right), a right preview card with a centered Open action and `Modified … · size · EXT` stats, and a static USEFUL INFO footer. Removes the old table/sort-header/kind-filter/pagination/group-toggle chrome and the conflicting legacy `.app .df-*` overrides in routines.css that were trumping the panel's own stylesheet via specificity. Preserves batch select/download/delete, rename, the row context menu, live-artifacts, plugin-folders, drag-upload, and the reloading overlay. Adds the designFiles.* section/kind/useful-info i18n keys across all locales. * feat: pick a working directory from Home before the project exists Let users choose a code folder on the Home screen up front, instead of only being able to replace it after a project is created. The HomeHero composer gains a folder pill (pick / clear); the chosen path is threaded through project creation and applied via the working-dir API once the project id exists. On desktop the working-dir POST is gated behind a host-minted token, so add a `dialog:pick-working-dir` IPC handler that shows the native picker and mints a single-use token bound to the folder, plus a `pickHostWorkingDir()` bridge + types/normalizer in `@open-design/host`. The renderer feature-detects the bridge and falls back to the web folder dialog (no token) otherwise. Native dialogs now allow creating a new folder inline (`createDirectory` / `ShowNewFolderButton`), and macOS uses `choose folder` so the panel reliably takes key focus. Wire the WorkingDirPill into the project composer footer and surface a reloading state + breadcrumb root label to the Design Files panel while a replace reindexes. Drop the WorkingDirPill "recent directories" localStorage list in favor of a single Replace action. * fix(web): preserve staged Home attachments when picking a working dir Move the working-dir handoff ahead of the initial upload so staged attachments land in the external folder instead of the temporary managed root, and fall back to the browser folder dialog when the desktop host build does not expose pickWorkingDir. * fix(web): surface host pick errors and normalize Windows dir basenames When the desktop host is present but cannot mint a working-dir token, surface the error instead of falling back to the browser folder dialog — the tokenless raw path would be rejected by the desktop auth gate as a late create-time failure. Only the pure web path (no token gate) keeps the dialog fallback. Also split the Home pill basename on both path separators so Windows paths show the folder name, not the full path. * fix(web,desktop): harden Home working-dir handoff against auth/token failures Run the desktop-auth handshake before minting the Home working-dir token so a missed startup registration fails up front instead of as a rejected deferred POST. Surface a create-time error toast when the working-dir handoff fails (expired ~60s token or daemon rejection) instead of silently leaving the project in the managed root while the user believes their folder was applied. * fix(web): abort create on failed working-dir handoff and always show root crumb When the pre-create working-dir handoff fails, short-circuit the rest of the Home submit path (skip the staged-attachment upload and auto-send) so the first run cannot proceed against the managed `.od/projects/<id>` tree the user did not choose, and reword the error toast accordingly. Always render the Design Files root breadcrumb, falling back to the `designFiles.crumbs` label when no rootDirName exists, so the toolbar is not blank on the default managed-storage view. Adds regressions for the short-circuit and default-root crumb cases. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) * fix(web): normalize Windows separators in WorkingDirPill basename shortPath() split only on '/', so a Windows working dir like C:\work\repo rendered as the full absolute path in the footer pill instead of the folder name. Split on both separators to match HomeHero/ProjectView, and add a regression covering the backslash path. Generated-By: looper 0.8.1 (runner=fixer, agent=claude-code) --------- Co-authored-by: qiongyu1999 <2694684348@qq.com> |
||
|
Harshit KhemaniandGitHub
|
fd8159216a |
feat(design-systems): import shadcn registry items as design systems (#3525)
* feat(design-systems): import shadcn registry items as design systems Add a third design-system import source alongside local folders and public GitHub repos: a shadcn registry item, referenced with the shadcn CLI shorthand `<owner>/<repo>/<item>` (resolved against the repo's root registry.json) or a direct https URL to a registry-item document. The item's cssVars theme is materialized into a temporary project and run through the existing local importer, so a shadcn theme yields the same DESIGN.md + tokens.css + manifest.json as every other import. Classic shadcn bare-HSL cssVars (e.g. "222 47% 11%") are wrapped into hsl(...) so brand colors survive instead of falling back to neutrals. Exposed on both surfaces, per the capability dual-track rule: - POST /api/design-systems/import/shadcn, with ImportShadcn* DTOs in packages/contracts and a "shadcn" source in the design-system manifest schema. - `od design-systems import-shadcn <reference> [--name --import-mode --craft --json]`. - Settings -> Design systems: a new "shadcn" import source (placeholder shadcn/ui/theme-zinc), with i18n keys added across all locales. http:// references are accepted only for loopback hosts so a self-hosted local registry can be imported and tests stay deterministic. * fix(design-systems): harden shadcn importer per review - Fail fast in writeShadcnFiles: a declared registry file that 404s, exceeds the size cap, exceeds the file-count cap, or has no safe destination now throws a path-named LocalDesignSystemImportError instead of silently producing a partially-materialized design system. - Resolve the shadcn `include` registry layout recursively; an item's relative file paths now resolve against the registry.json that declared it (so `owner/repo/item` works for registries that use include). - Keep the fetch abort timer armed across response-body consumption so a server that stalls the body cannot hang the import. - Block SSRF: redirect:'error' on the default fetch, plus per-fetch host re-validation (https, or http only for loopback) on every URL we contact - entry point, raw GitHub URLs, include targets, and raw files. Adds regression tests for include resolution, a 404 declared file, an oversized inline file, and a non-loopback http include target. * fix(design-systems): close two shadcn importer review gaps - parseShadcnReference: wrap decodeURIComponent of the URL fragment so malformed percent-encoding (e.g. "#%E0%A4%A") surfaces as a 400 BAD_REQUEST instead of an uncaught 500 from the import route. - writeShadcnFiles: two declared files that collapse to the same sanitized destination (e.g. "@/button.tsx" + "button.tsx") now fail as a BAD_REQUEST naming both sources, instead of silently dropping the later file - the same partial-materialization failure mode this PR avoids. Adds parser + importer regression tests for both. * fix(design-systems): harden shadcn importer (SSRF, resource bounds, fidelity) A deeper hardening pass from an adversarial self-review of the importer. SSRF egress policy (assertFetchableUrl): loopback is allowed over http or https (self-hosted/local registries); private (10/8, 172.16/12, 192.168/16, 100.64/10), link-local (169.254/16, fe80::/10), ULA (fc00::/7), unspecified (0.0.0.0, ::), and IPv4-mapped/integer-encoded internal addresses are refused for BOTH schemes; any other host must be https. Enforced at parse time and on every fetch - entry point, include targets, and raw files - so a direct https://169.254.169.254, or an include/redirect pivot to an internal address, is blocked rather than only the http downgrade. Resource bounds: bodies are read with a running byte cap (no unbounded buffering) plus a Content-Length pre-check; the whole import shares one wall-clock deadline and a max-total-requests budget; include breadth and file count are capped. A stalled/huge body or a large include fan-out can no longer hang or OOM the import. Fidelity / fail-fast: a files[] entry with no usable destination is now a 400 (was silently skipped); registry files are materialized under a subdirectory so a crafted target cannot overwrite the importer's theme.css/package.json/ README.md; cssVars keys/values are validated so a hostile entry cannot inject extra declarations or be silently dropped; slashed git refs are encoded per segment; direct-URL items resolve path-only files relative to the item URL; fetch timeouts surface as a clean BAD_REQUEST. Adds tests for each: https-internal SSRF (entry + include), a no-destination file, a reserved-file overwrite attempt, a slashed ref, a direct-URL path-only file, cssVars injection, and a streaming over-cap body. |
||
|
|
b2a9ebfa09 |
feat(landing-page): Solution nav + six use-case pages (real catalogue examples, responsive, 18 locales) (#3598)
* feat(landing-page): restructure header into Solution / Agent / Resources dropdowns The top nav had no entry point for the highest-leverage SEO surfaces. Pages that already exist — /compare/, /alternatives/claude-design/, /agents/ — were unreachable from the bar, and there was nowhere to grow the planned Claude Code / Figma / Lovable / use-case pages into. This adds three dropdowns and folds the standalone Blog + Tutorials links into a Resources group so the bar still fits at the desktop breakpoint: - Solution: Claude Code for Design, AI Prototype / Dashboard / Landing Page / Slides generators, Figma / Lovable alternatives. Sub-item names stay English (each is a target search phrase). Backlog pages point at an existing surface today and get repointed once their dedicated page ships. - Agent: the 17 first-party coding-agent adapters, each deep-linking to its anchor on the /agents/ hub. Names mirror the canonical defs on that page. - Resources: Blog, Tutorials, Compare, Download. i18n: solution / agent / resources nav labels added to all 18 locales and the HeaderCopy.nav type. Sub-item names are not localized by design. No CSS changes — the existing .nav-dropdown / mobile-flatten rules cover the new dropdowns. Existing GSC-ranked routes (/plugins/skills, /html-anything) are untouched. * fix(landing-page): align header dropdowns 1:1 with the agreed nav spec The first pass restructured the bar around an SEO-keyword grouping that diverged from the agreed Header spec. This rewrites the dropdowns to mirror the spec exactly: - Product: OpenDesign, html-anything (AMR removed — it now heads Agent). - Solution: two labeled groups — Use cases (Prototype, Dashboard, Slides, Image, Video, Design System) and Roles (Solo Builder, Designer, Engineering, Product Managers, Marketing). Sub-item hrefs are placeholders until each dedicated page ships. - Agent: AMR (the design Agent) followed by the 17 coding-agent adapters, named verbatim per the spec (Codex CLI, Qwen Code, Grok Build, …), each deep-linking to its /agents/ hub anchor. - Plugins: Design Templates / Design Skills / Design Systems. - Resources: Blog, Video Tutorials, Weekly Newsletter (disabled placeholder), Download. - Community: promoted to a dropdown — Contributors, Ambassadors, Moderators, Discord, Discussions. i18n: solution / agent / resources / useCases / roles nav labels across all 18 locales. Sub-item names stay English by design. globals.css gains group labels, a disabled-item style, a 2-column Agent menu, and a height cap on the tall Solution menu. * feat(landing-page): add /solutions/prototype/ use-case page First of the Solution use-case pages, and the template the other ten (5 use cases + 5 roles) will copy. Explains how a coding agent uses Open Design to turn a prompt into a clickable, multi-screen prototype — image + text + table, the format the nav doc calls for. - Page body: hero illustration, TL;DR, a four-step "how it works" list, a capability table (Open Design vs. traditional prototyping tools), a worked example linking to /plugins/templates/, FAQ (FAQPage schema), and CTA. - Copy lives in a new self-contained app/solution-pages-i18n.ts (English + hand-reviewed Simplified Chinese; other locales fall back to English). Kept out of the big InfoPageCopy shape so rich page bodies don't force a hand-written entry per locale across all 18 compact locales. - Illustrations are AI-generated, warm-paper editorial style, hosted on the marketing R2 bucket under landing/assets/solutions/prototype/ and served through the existing cdn-cgi/image pipeline. - New CSS in sub-pages.css for the hero/figure/stepped-how-to blocks; reuses .compare-table / .tldr-card / .faq-list / .info-cta. - Header: the Solution dropdown's Prototype item now points at the real page. - Adds the root page + the [locale] wrapper so all 18 locales build. * feat(landing-page): add the five remaining Solution use-case pages Completes the Solution → Use cases set (Prototype was the first; this adds Dashboard, Slides, Image, Video, and Design System). Each page explains how a coding agent uses Open Design for that workflow — image + text + table, the format the nav doc calls for: - Warm-paper editorial hero illustration (AI-generated, on the marketing R2 bucket under landing/assets/solutions/<slug>/). - TL;DR, a stepped "how it works" list, and a capability table contrasting Open Design with the traditional tool for that job. - A "template gallery" section showing REAL in-repo example thumbnails for that scenario (from public/previews/plugins/) linking to /plugins/, so the examples are genuine outputs rather than mockups. - FAQ (FAQPage schema) and a CTA. Copy lives in app/solution-pages-i18n.ts as self-contained English + hand-reviewed Simplified Chinese blocks; other locales fall back to English. The shape gained a `gallery` field, and the prototype page was migrated from its single worked-example image to the same gallery treatment so all six pages are structurally identical. Header: the Solution dropdown's six use-case items now point at their real pages. New gallery + grid CSS in sub-pages.css. Each page ships a root + [locale] wrapper so all 18 locales build. Agent detail pages and the Role pages are intentionally deferred to a follow-up PR to keep this one reviewable. * feat(landing-page): real catalogue examples, wide responsive layout, full i18n for Solution pages Reworks the six Solution use-case pages and the nav off review feedback, and translates them into all 18 locales. Pages (prototype / dashboard / slides / image / video / design-system): - Wider, Figma-style layout scoped to `.solution-page` (max-width 1180, a two-column hero band, tighter section rhythm, 880/560 breakpoints) — the other info-pages keep their 920px measure untouched. - Example sections now pull REAL catalogue records instead of editorial mockups: image/video/slides render the matching Image / Video / deck plugins via <TemplateCard> (video plays its Cloudflare Stream clip on hover); design-system renders real palette swatch cards via <SystemCard>. Cards hand-screened for strong covers and laid out 3-up. - Feature grid + template gallery use distinct images (no intra-page repeats). - Hero secondary CTA points at /download/; the footer CTA drops Quickstart. Nav (header.tsx): - Agent dropdown trimmed to AMR only (the 17 coding-agent rows kept in a commented block for a later pass). - Solution dropdown's Roles group hidden until those pages exist. - Use-case sub-items read each page's localized breadcrumb; Plugins / Resources / Community sub-items use localized nav copy. Weekly Newsletter removed. i18n: - solution-pages-i18n.ts gains 16 machine-translated locale blocks (ja, ko, de, fr, ru, es, pt-br, it, vi, pl, id, nl, ar, tr, uk, zh-tw) on top of the hand-written en + zh; BY_LOCALE wires all 18. Brand names, thumbs and hrefs left verbatim. - HeaderCopy.nav gains solution/agent/resources/useCases/roles/contributors/ ambassadors/moderators across all 18 locales. * fix(landing-page): address review on Solution pages (real previews, localized JSON-LD, crash-visible selection) - prototype/dashboard feature grids now render real catalogue records via <TemplateCard> (prototype-category plugins) instead of `page.features` thumbs. The old example-* basenames resolved only because the gitignored public/previews/ dir happened to hold them locally; the grids now point at pipeline-emitted /previews/plugins/<manifest-id>.png like image/video/slides. - All six pages emit locale-aware JSON-LD URLs (BreadcrumbList + WebPage) built from the request path, so /ja/, /de/, … no longer ship the English canonical. - Featured-record selection is crash-visible: new pickSolutionPlugins() (and an inline guard on design-system) throws at build if any pinned manifest id is missing, instead of silently dropping it via .filter(Boolean) — per AGENTS.md. (This caught example-live-dashboard, which categorizes as live-artifact, not prototype; swapped to example-last30days.) - Removed the dead AGENT_LINKS constant from header.tsx. * fix(landing-page): drop dead nav-dropdown CSS left after the Agent/Newsletter trim `.nav-dropdown .dropdown-disabled` (Weekly Newsletter placeholder) and `.nav-dropdown-agents` (the 2-column 17-agent layout) are unreferenced now that the Newsletter item is gone and the Agent dropdown lists only AMR. Removed both; `.nav-dropdown-solution` stays (still used by the Solution dropdown). --------- Co-authored-by: Joey-nexu <joeylee12629@gmail.com> |
||
|
|
1e1cf78f8f |
[CC] feat(settings): re-add Max Tokens field to BYOK settings (#3348)
The per-request output cap input was dropped from the API/BYOK execution section in #250 (split API provider protocols). The supporting state (`AppConfig.maxTokens`, `effectiveMaxTokens`, `MIN/MAX_MAX_TOKENS`, `modelMaxTokensDefault`) and the i18n strings (`settings.maxTokens` / `settings.maxTokensHint`) all survived, but the UI control did not — the constants stayed imported in SettingsDialog yet unused, so users had no way to raise the cap without editing localStorage directly. This restores the field after the BYOK base-URL input. With no override, `effectiveMaxTokens` falls back to the per-model default and ultimately `FALLBACK_MAX_TOKENS` (8192); provider/model ids that miss the lookup table (e.g. PPIO's `pa/`-prefixed Claude ids) silently clip long generations at 8192 with no way to lift it from the UI. Re-adding the input lets BYOK users set a higher cap (bounded to [1024, 200000]); a valid override wins over the fallback in `effectiveMaxTokens`. Co-authored-by: superguo999 <81609822+superguo999@users.noreply.github.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
|
a6cfc1c175 |
fix(web): keep the draw toolbar usable in tablet/mobile preview (#3459)
* fix(web): keep the draw toolbar usable in tablet/mobile preview In tablet/mobile preview the draw annotation toolbar was rendered inside the scaled, overflow-hidden device-frame shell (width = --preview-viewport-width, e.g. 390px on mobile), so the toolbar — wider than the frame — overflowed and its right edge (the Queue button) was clipped and unusable. Portal the toolbar (and the capture-warning) to the non-scrolling preview body (.viewer-body) so it stays fully visible and pinned regardless of the selected device viewport, instead of being clipped by the frame or scrolling away with the preview content. The ink canvas stays inside the frame, aligned to the iframe; it falls back to inline rendering when there is no .viewer-body ancestor. To stop the floating toolbar from hiding the bottom of the page and its scrollbar, reserve room at the bottom of the scroll area while drawing in a tablet/mobile frame so the device frame can clear the toolbar, keeping the checkered backdrop full-bleed (a floating pill, not a docked footer bar). Adds a regression test that fails on main (toolbar nested inside the clipped wrap) and passes with the portal. Fixes #3455 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(web): resolve the draw-toolbar host pre-paint to avoid a one-frame flash Use useLayoutEffect (not useEffect) to look up the .viewer-body portal host so the first active paint is already portaled. With the post-paint effect the clipped inline toolbar could flash for one frame when entering annotate mode in a tablet/mobile device frame. Addresses review feedback on #3459. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Siri-Ray <2667192167@qq.com> |
||
|
|
33d59ebb3a |
Clarify OpenCode reasoning support and avoid unsupported reasoning argv (#3346)
* Clarify OpenCode reasoning support * Address PR review suggestions --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> |
||
|
yinjialuandGitHub
|
836d734455 |
[codex] Add same-run retry runtime (#3576)
* Add same-run retry runtime * fix(daemon): route retried-start and watchdog stall through retry finalizer Two retryable failure paths bypassed the same-run retry finalizer, so they emitted no run_retry_attempted/run_retry_finished telemetry and could report the fallback retry_final_result: 'not_attempted'. - restartSameRunAfterRetry's catch now routes a failed retried start through finishWithRetryDecision instead of design.runs.finish, so it emits terminal retry telemetry and sets run.retryFinalResult. retryAttemptCount is already 1 there, so decideSafeRunRetry suppresses with attempt_limit_reached and cannot loop. - failForInactivity's no-output stall now routes through the finalizer after surfacing stallPayload. A silent first-token hang classifies as a retryable timeout at first_token_wait, so the same-run retry now recovers it. A new watchdogRetryRestarted guard makes the SIGTERM'd stalled child's close bail early (revoking only its own tool token) so it cannot re-finalize the run or delete the new attempt's event sink (both keyed by the shared runId). - finishWithRetryDecision returns whether it restarted so the watchdog can set the guard. Adds a run-retry-runtime test with a low OD_CHAT_RUN_INACTIVITY_TIMEOUT_MS proving the same-run retry fires for a silent first attempt. |
||
|
|
dbc74a397c |
feat(agents): CLI detection self-rescue with actionable diagnostics (#3560)
* feat(agents): surface actionable CLI diagnostics Detect unavailable CLI agents with structured diagnostics, stream agent discovery results to the settings UI, and render actionable fix controls for common setup failures. Also harden the streamed agents client so server error events and dropped streams surface as failures instead of replacing state with a truncated agent list. * fix(agents): gate streamed rescan loading Generated-By: looper 0.9.2 (runner=fixer, agent=codex) * fix(agents): distinguish non-executable CLI diagnostics Generated-By: looper 0.9.2 (runner=fixer, agent=codex) * fix(agents): defer OAuth diagnostics without producer Generated-By: looper 0.9.2 (runner=fixer, agent=codex) --------- Co-authored-by: chaoxiaoche <chaoxiaoche@chaoxiaochedeMacBook-Pro.local> Co-authored-by: Looper <looper@noreply.github.com> Co-authored-by: Looper Fixer <looper-fixer@noreply.github.com> Co-authored-by: looper-fixer <looper-fixer@users.noreply.github.com> |
||
|
|
f3ec4968af |
feat(web): revamp chat UI styling + live streaming code card (#3382)
* feat(web): revamp chat UI styling + live streaming code card Restyle the chat pane (chat/tokens/code/tools/routines/composio CSS) and add a live code-component card that streams Write/Edit tool input and <artifact type="text/html"> bodies character-by-character via a new ephemeral tool_input_delta SSE event, flipping to the completed FileWriteCard/FileEditCard view on close. Includes shiki highlighting, markdown/artifact strip handling, tool-status cleanup, and i18n keys. * chore(nix): refresh pnpm deps hash for merged lock --------- Co-authored-by: qiongyu1999 <2694684348@qq.com> |
||
|
|
90aab29695 |
feat(analytics): attribute AMR entry conversions (#3579)
* feat(analytics): attribute AMR entry conversions
* test(web): cover AMR attribution entry points
* fix(daemon): gate AMR entry mirror on analytics consent
The /api/integrations/vela/analytics-entry route forwarded every AMR
entry click to the external AMR ingest endpoint regardless of consent:
mirrorAmrEntryAnalytics fired whenever globalThis.fetch existed, so an
opted-out user's click was still shipped to a third-party service.
Gate the mirror the same way the analytics-capture path does:
- null readAnalyticsContext (no x-od-analytics-* headers => Privacy
metrics off) short-circuits to { mirrored: false }.
- defense-in-depth re-read of telemetry.metrics from app-config drops
the mirror even if a stale header leak slips through after opt-out.
Adds two daemon route red specs asserting that a no-headers request and
a metrics-off request never hit the capture server.
---------
Co-authored-by: lefarcen <935902669@qq.com>
|