mirror of
https://github.com/CopilotKit/CopilotKit.git
synced 2026-09-14 16:26:20 +08:00
Jordan Ritter
87b369fdb4
## Summary - Reworks the `declarative-gen-ui` demo (LangGraph Python + Google ADK) to feel like Beautiful Chat's sales dashboard, per [OSS-136](https://linear.app/copilotkit/issue/OSS-136/declarative-gen-ui-demo-rework-to-feel-like-beautiful-chats-sales) - Suggestion pills are natural business questions; chart-type steering moved from user prompts into the agent system prompt + frontend context (`sales-context.ts`, shared verbatim by both integrations) - Every pill renders a dashboard-grade surface: - **Show my sales dashboard** — bare KPI strip + regional revenue donut + 6-month revenue bars (no surrounding card) - **Team performance** — rep table + quota-attainment bar chart - **Anything at risk?** — risk KPI strip over three severity cards (icon badges, reason + next action) - **Top account details** — account fact card + product-line donut - Renderers ported to beautiful-chat's visual language: card chrome, metric typography, recharts donut/bars, shared palette, lucide severity icons - Test triad synced: D5 probe (per-pill newly-mounted testid assertions), Playwright e2e (incl. click-dispatch guard), aimock fixtures **captured from live model responses**, QA docs ## Verification - Fixture validation: 738/738 · harness probe unit tests: 17/17 - Container e2e (fixture replay, both integrations rebuilt): LGP 6/6 · ADK 6/6 - Live-LLM iteration on LGP: all four pills produce the steered composition consistently (11/11 captured runs + repeated interactive verification) ## Test plan - [ ] `pnpm exec nx run @copilotkit/showcase-harness:test -- d5-gen-ui-declarative` - [ ] `pnpm --filter @copilotkit/showcase-scripts test aimock-fixtures` - [ ] `BASE_URL=<container> npx playwright test declarative-gen-ui.spec.ts` per integration - [ ] Post-deploy: https://dashboard.showcase.copilotkit.ai/#matrix:links,health row "Declarative UI: Dynamic A2UI" still reaches D5 (CV badge) for langgraph-python and google-adk