Codex
34 lines
1.8 KiB
Markdown
34 lines
1.8 KiB
Markdown
# Product
|
|
|
|
## Register
|
|
|
|
product
|
|
|
|
## Users
|
|
|
|
SignalScout is used by one operator who wants a daily AI-generated intelligence brief without manually searching across news, X, GitHub, and company announcements. The user is usually checking what changed, which projects matter, and which links deserve follow-up.
|
|
|
|
## Product Purpose
|
|
|
|
The product searches dynamic news candidates and recent high-heat GitHub project candidates, then uses Grok to structure AI news, model releases, agent framework movement, product launches, and funding signals. Success means the user can open one dashboard, see concrete collected items with sources and context, inspect the latest daily report, and trigger a fresh collection run when needed.
|
|
|
|
## Brand Personality
|
|
|
|
Quiet, sharp, operational. The interface should feel like an intelligence desk rather than a marketing site.
|
|
|
|
## Anti-references
|
|
|
|
Do not make a landing page, generic SaaS hero, decorative card grid, or empty analytics dashboard. Avoid fake metrics that do not come from collected data. Avoid loud neon AI styling, glass panels, gradient text, and ornamental charts.
|
|
|
|
## Design Principles
|
|
|
|
- Evidence first: every important item should expose its source and enough context to verify it.
|
|
- Workbench over brochure: the first screen should help the user read, filter, and act.
|
|
- Dense but calm: support repeated daily review without visual noise.
|
|
- Agent output is inspectable: show summaries, entities, run state, and report text clearly.
|
|
- One path: the UI reflects the current News Search + GitHub + Grok + MySQL pipeline.
|
|
|
|
## Accessibility & Inclusion
|
|
|
|
Use semantic controls, visible focus states, sufficient contrast, reduced-motion friendly transitions, and readable Chinese/English mixed content. Target practical WCAG AA behavior for this private dashboard.
|