@orkestrel/rater 0.0.1 → 0.0.3

package/README.md

@@ -1,18 +1,13 @@
 # @orkestrel/rater
 
-A typed **rating engine**: programs compile line, pass, and ruling definitions
-into determinations, worksheets, and aggregates for rated subjects. Pure,
-authored **program definitions** — `lines` (quantitative rating), `passes`
-(ordered pre-rating overlays), `rulings` (rule id → effect routing), `notices`,
-and an optional `authority` (a final logical gate deriving a `decision`) — are
-compiled and rated over ONE shared [`@orkestrel/reason`](https://github.com/orkestrel/reason)
-engine (`quantitative` + `logical` reasoners). Rating never mutates its
-inputs: every result — `LineResult`, `ProgramResult`, `SubjectResult`,
-`AggregateResult` — is a fresh object carrying a `trace` and accumulated
-`errors`. Batch rating additionally supports per-program aggregate fields, an
-optional partition key, and gated aggregate determinations, plus per-`Status`
-tallies. Environment-agnostic — no I/O, no browser or server assumptions.
-Part of the `@orkestrel` line.
+A typed **quantitative rating layer** over [`@orkestrel/reason`](https://github.com/orkestrel/reason):
+authored **lines** — each a plain reason `QuantitativeDefinition` joined to display
+metadata — are rated against a **subject** (a plain data record) to produce a
+`LineResult` per line (an `amount` plus its `Worksheet` audit trail) and one
+`RatingResult` (every line's outcome plus a derived `total`). The caller decides
+which lines to rate; `Rater` only evaluates what it is given. Rating never mutates
+its inputs — every result is a fresh object. Environment-agnostic — no I/O, no
+browser or server assumptions. Part of the `@orkestrel` line.
 
 ## Install
 
@@ -23,12 +18,12 @@ npm install @orkestrel/rater
 ## Requirements
 
 - Node.js >= 24
-- ESM-only (no CommonJS build)
+- ESM (`import`) and CommonJS (`require`) via the `exports` field
 
 ## Usage
 
 ```ts
-import { createRater, lineDefinition, programDefinition } from '@orkestrel/rater'
+import { createRater, lineDefinition } from '@orkestrel/rater'
 import { factorGroup, quantitativeDefinition, staticFactor } from '@orkestrel/reason'
 
 const rater = createRater()
@@ -41,26 +36,23 @@ const base = lineDefinition(
 	]),
 )
 
-rater.programs.add(programDefinition('p1', 'Program', [base]))
+const result = rater.rate([base], { id: 'subject-1' })
+result.lines[0]?.amount // 100
+result.total // 100
 
-const result = rater.rate({ id: 'subject-1' }) // one subject → one SubjectResult
-result.programs[0]?.eligibility // 'eligible'
-
-rater.rate([{ id: 'a' }, { id: 'b' }]) // an ARRAY resolves to the batch AggregateResult overload
+rater.emitter.on('rate', (subject, rated) => rated.success)
 
 rater.destroy()
 ```
 
-`rate` dispatches by input shape — an ARRAY resolves to the batch
-`AggregateResult` overload, a single subject resolves to `SubjectResult`.
-Every applied determination and derived decision fires through
-`rater.emitter` (`rate` / `aggregate` / `determine` / `decide`).
+`rate` accepts a plain `LineDefinition[]` or a full `RatingDefinition` plus one
+subject — both overloads rate a single subject. Every `rate` call fires once
+through `rater.emitter` (`rate`).
 
 ## Guide
 
-For the full surface — the `Rater` orchestrator, compiled `Program`s, the
-ordered `ProgramManager`, worksheet and determination helpers, validators,
-factories, errors, and the observation surface — see
+For the full surface — `Rater`, `RatingResult`, worksheet types, validators,
+factories, errors, and options — see
 [`guides/src/rater.md`](guides/src/rater.md).
 
 ## Package