Design system · built from 78 answered questions
Dense data
Compact, exact and quiet, for analysts who scan alerts and compare signals all day.
58components
172primitives
2themes
| Alert | Asset | Detected | Risk score | Status |
|---|---|---|---|---|
| ALR-2041 | api-gateway-01 | 08:30 | 92 | Resolved |
| ALR-2042 | llm-inference-02 | 09:10 | 78 | Escalated |
| ALR-2043 | vpn-edge-03 | 09:45 | 99 | Breach confirmed |
| ALR-2044 | model-registry | 10:20 | 31 | False positive |
| ALR-2045 | auth-service | 11:00 | 64 | Investigating |
| ALR-2046 | db-primary | 11:30 | 81 | Escalated |
Why it looks like this
Every decision follows from an answer. These are the answers its components and rules cite most. Follow them all in the Decision reference.
What is the main content?
Numbers and records
Most of the screen is numbers and records.
content.q1How much is on one screen?
A lot
Experts want many records on one screen.
content.q2What is a costly mistake?
Acting on the wrong record
Acting on the wrong row is the costly error.
outcomes.q2Explore the system
PrimitivesAll 172 raw values, and the tokens that point to each.
Aa
FoundationsColor, type, spacing, shape and motion, each with the answer behind it.
InvestigateEscalated
Components58 components, each with when to use it and how Dense data decided.
Lead with the object, then the state: "Alert 2041 contained."
Voice and toneTerse and exact. Say what happened and what to do, in as few words as possible.
Themes2 themes. Only semantic tokens change; components never do.
open-ds why dense-data buttonFor agentsAGENTS.md, the skill and the MCP: an AI builds the way the system would.
Who it is for
- Use it for: Security operations, threat detection, model monitoring, admin panels and analytics: many records, daily expert users, mouse and keyboard.
- Not for: Public sites, reading products, or first-time visitors on phones.
- Products like: Security operations console, Threat detection, AI model monitoring, Analytics dashboard
- Themes: light, dark
Use scenarios
- A security analyst opens the alert queue at 8:00 and scans 200 rows for escalated threats.
- An analyst compares this week's detections with last week's, model by model and column by column.
Every answer
All 78 answers, and the design decisions each one made: Decision reference.