Docs
Docs
Pressure is one reading per dataset per UTC day: its state, one of five from the least pressure to the most (Expanded, Released, Equilibrium, Pressurized, Compressed), plus its flow: which way it is moving that day (building or releasing) and whether that movement is accelerating or slowing. Two ways to read it, each with its own page: MCP for AI assistants and agents, and the REST API for curl, scripts, and jobs. Both serve the same data from the same pipeline; what they share is below.
Freshness
- Pressure is daily: one reading per dataset per UTC day, final once the UTC day closes.
- A new account's readings end a week before the latest reading; the latest flow is on the home page.
- Responses are cached: about five minutes while a window includes the still-settling day, up to a day once every day in it has closed. A repeated call inside that window costs nothing new.
Metrics
Every dataset serves one daily reading: a state, named by its label (as
Pressure is), or a value on its own scale with a label for the range it falls
in. Where the metric carries one, it also has a flow: which
way it is moving (building or releasing) and whether that movement is
accelerating or slowing. The table lists what an account can read today;
it follows the live catalog, and list_datasets returns the same
list with your own access. A metric's explanation, when its editors wrote
one, appears under its name.
| Metric | Dataset | Scale | Labels | Included |
|---|---|---|---|---|
| Pressure | Macro (Global) | 5 states | expanded, released, equilibrium, pressurized, compressed; flow: building or releasing, pace accelerating or slowing | 90 days, ending a week back |
Showing the launch catalog while the live list loads.
Limits
- 200 calls a day, 10 a minute, per account.
- MCP and REST draw from the same budget.