Dashboard design improvements - #42
Merged
Merged
Conversation
…izing/labels, add chart expand - DashboardWorkspace/DashboardsHome: translucent chrome, press feedback, spring-flavored hover/entrance transitions, reduced-motion and reduced-transparency fallbacks, refined chat trace panel with a step timeline. - DashboardArtifact: fade artifact in on load; fix a feedback loop where a CSS height transition plus a stale interval poll made dashboard height grow continuously on any reflow (e.g. page scroll). - dashboardChartLib: cap chart height regardless of container width (was inflating on wide cards); add a built-in per-chart expand-to-fullscreen control; darken/enlarge and truncate x-axis labels so long names no longer render illegible or overlap their neighbor. - dashboard-design SKILL.md: document the automatic expand control, add a named "avoid AI-slop patterns" section (gradients, emoji icons, decorative blobs, off-scale spacing/type, chart color discipline) wired into the existing self-review checklist; bump skill version to 2.1.0.
…rd generation Every message sent in the dashboard builder's chat — including a bare "hi" — was wrapped in DashboardAgentService.buildTask's full ground+write-SQL+verify+ self-review pipeline, since there was no other path. That's why trivial messages took ~3 minutes: the agent was dutifully trying to build a dashboard in response to a greeting. DashboardAgentService now classifies each message with a direct, single ChatModel call (no agent session, no tools) before deciding which path to take — a real LLM judgment call, not a keyword match, since a fixed word list can't tell "make it prettier" or "no, the other one" from a real edit ask. The classification prompt is explicitly biased toward "this is a real request" on anything ambiguous, so a genuine request is never misrouted to the lightweight reply-only path. Verified against the running backend: "hi" and "thanks!" both classify as chat-only and reply in seconds; "make it prettier" (which a keyword list would have missed — no build-related word in it) correctly routes to the full grounding+SQL+verify pipeline. Frontend (dashboardGenerator.js, DashboardWorkspace.jsx) grows an onChat handler that appends the reply as a plain agent bubble without touching the saved config or triggering an auto-save.
4 tasks
5 tasks
venkateshsakamuri-lab
added a commit
that referenced
this pull request
Aug 13, 2026
…builds (#46) ## Summary - Backend chat-only replies (e.g. `"hi"`) now stream as a dedicated SSE `chat` event instead of reusing `done`. - Frontend handles `chat` separately and guards the workspace `onDone` path so a chat-shaped payload cannot append the canned “Done — built…” save message or clobber the artifact. ## Why Sumit’s chat-intent classifier in #42 was working (Hermes ran the greeting task), but the UI still showed “Done — built and verified…” because that line is hardcoded on the build-success path, which was driven by the shared `done` event. ## Test plan - [x] Hard-refresh Dashboards, open an existing draft, send `hi` - [x] Expect a normal greeting reply — not “Done — built…” - [x] Artifact HTML / KPIs unchanged after the greeting - [x] A real edit (“add a filter”) still streams `done` and saves as before - [x] `curl`/SSE: `prompt=hi` yields `event:chat` and no `event:done`
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.