> ## Documentation Index
> Fetch the complete documentation index at: https://docs-attestly.code4source.com/llms.txt
> Use this file to discover all available pages before exploring further.

# How it works

> Geometry → sources → checks → verdict.

Every call follows the same flow:

```
your geometry  ─┐
                ├─→  match against reference sources
your ruleset   ─┘    (protected areas, sanctions, registers, ...)
                          │
                          ▼
                    run checks (count, area, distance, subject match, ...)
                          │
                          ▼
                    compose a verdict
```

## The four primitives

### Sets

A **set** is a collection of features (polygons, points, lines, or
subject records). Sets are either anchored on a source — e.g. "all
features in `funai_tis` that intersect my input
geometry" — or built from set operations (`union`, `intersection`,
`difference`) over other sets.

### Projections

A **projection** computes a value from a set: `count`,
`total_area_m2`, `min_distance_m`, an aggregate over a feature
property, or a merge of geometries. Projections produce numbers and
geometries — they do not contribute to the verdict on their own.

### Checks

A **check** verifies something about a set or a projection and
produces a boolean outcome with a severity. Two flavors:

* `exists` — is the set non-empty?
* `threshold` — does a projection's value pass an `op` + `value`
  comparison? (e.g. `count >= 1`, `total_area_m2 > 50000`)

### Verdict

The verdict combines check outcomes by severity. A response carries
two enums on purpose:

* `status` — `ok` or `degraded`. **Execution health.**
* `outcome` — `compliant`, `warning`, `non_compliant`, or `degraded`.
  **Compliance result.**

See [Verdicts](/concepts/verdicts) for the full semantics.

## Three styles of usage

| Style           | Endpoint                            | When to use                                   |
| --------------- | ----------------------------------- | --------------------------------------------- |
| Raw discovery   | `/v1/intersections`, `/v1/distance` | Exploration, debugging, ad-hoc queries        |
| Subject lookup  | `/v1/subjects`                      | Sanctions / blacklist / registry checks by ID |
| Full compliance | `/v1/evaluate`                      | Production decisions against a ruleset        |

Most customers start with discovery, then move to `/v1/evaluate` once
their compliance policy stabilizes.