zadao.aiTrending Tools & Usage Stats
See which Nova tools the community is actually using. The Tools page now shows a live Trending Tools panel — a ranked, real-time view of the most-used tools across both the REST API and the MCP server.
In short
The Trending Tools panel ranks the most-used Nova tools by week, month or all time, with trend arrows and the REST vs MCP split. Aggregate only; also available at GET /api/tools/stats.
What You'll See
| Element | Description |
|---|---|
| Ranked tools | The most-used tools for the selected period, with an animated usage bar |
| Time range | Toggle between Week, Month, and All-time |
| Trend arrows | Week-over-week change (▲ up / ▼ down) vs the previous equal-length period |
| REST · MCP split | What share of each tool's usage came via the MCP server vs the REST API |
| Headline totals | Total runs, unique callers, and success rate for the period |
| Daily sparkline | A compact chart of daily activity across the range |
| Card badges | Tools are flagged ★ Most used or 🔥 Trending directly on their cards |
How Trending Is Calculated
- Most used — ranked purely by number of successful runs in the selected period.
- Trending — a tool whose usage grew meaningfully versus the previous equal-length period (e.g. this week vs last week).
- To keep things honest, a tool needs a minimum volume before a trend percentage is shown — so a tool going from 1 → 3 calls won't be labelled "+200% 🔥". Low-traffic tools simply show no arrow.
- All-time has no trend arrows (there's no prior period to compare against).
Privacy
The public trending view is aggregate only. It never exposes who called a tool:
- ✅ Shown: tool name, run counts, REST/MCP split, trends, success rate, unique-caller count.
- ❌ Never shown publicly: caller wallet addresses, IP addresses, or revenue.
Caller-level detail and revenue are visible only to platform admins in the internal dashboard.
For Builders
Want the raw numbers? The same aggregate data powers a public endpoint:
GET /api/tools/stats?range=week|month|all
It returns ranked per-tool usage, headline totals, and a daily series — all aggregate, no identities. Responses are cached for a few minutes. The MCP vs REST split is the quickest way to see what agents lean on versus what people run from the dashboard.