brepjs-verify 0.2.1 → 0.3.0

package/CHANGELOG.md

@@ -1,5 +1,18 @@
 # Changelog
 
+## [0.3.0](https://github.com/andymai/brepjs/compare/brepjs-verify-v0.2.1...brepjs-verify-v0.3.0) (2026-06-04)
+
+
+### Features
+
+* **brepjs-verify:** live text-to-cad eval flywheel ([#1215](https://github.com/andymai/brepjs/issues/1215)) ([4e81fc4](https://github.com/andymai/brepjs/commit/4e81fc4053491ce3e08182d57d76bd649252ea3c))
+* **brepjs-verify:** standalone bundled CLI + rename from brepjs-cad ([#1211](https://github.com/andymai/brepjs/issues/1211)) ([05b3799](https://github.com/andymai/brepjs/commit/05b3799a0e9ee4968d4cac92f3a2ea236e39cd35))
+
+
+### Bug Fixes
+
+* **brepjs-verify:** correct fillet/chamfer arg order in no-edges hints ([#1218](https://github.com/andymai/brepjs/issues/1218)) ([835f13a](https://github.com/andymai/brepjs/commit/835f13ac966b4264ba56a5cfc371bbbbbd1a0f01))
+
 ## [0.2.1](https://github.com/andymai/brepjs/compare/brepjs-cad-v0.2.0...brepjs-cad-v0.2.1) (2026-06-04)
 
 

package/README.md

@@ -71,6 +71,23 @@ Few-shot examples live under `skill/examples/<name>.brep.ts`, each with a `<name
 
 `npm run eval` (`bench/run.ts`) replays every `skill/examples/*.brep.ts` with a sibling `*.expected.json` through the public `runPart` runtime, compares measured volume/area/validity/shape-type against the recorded baseline within each file's tolerance (default 0.5%), prints a PASS/FAIL scorecard, and exits non-zero on any regression. It is deterministic — no LLM or API key — so it runs in CI as the package's regression net. Refresh a baseline by re-recording the example's `*.expected.json` after an intentional geometry change.
 
+### Live eval (`npm run eval:live`)
+
+The measurement flywheel: sends ~18 natural-language part prompts (`bench/prompts.ts`) to a real model — using the **deployed `SKILL.md` as the system prompt**, so it measures the actual skill — then verifies each generated part two ways:
+
+- **Auto (objective):** `runPart --check` → valid solid + any pinned dims within tolerance.
+- **Judge (intent):** a multimodal Claude call looks at the rendered iso/front/top/right snapshots and decides whether the part matches the request + rubric.
+
+The scorecard reports per-category `valid` / `judge` / `both` rates and stamps the model + **resolved brepjs version** + date (so trend lines don't mix kernel versions).
+
+```bash
+ANTHROPIC_API_KEY=sk-... npm run eval:live -w brepjs-verify          # opus by default
+ANTHROPIC_API_KEY=sk-... npm run eval:live -w brepjs-verify -- --model claude-sonnet-4-6
+#   --only <id|category>   run a subset      --keep   keep the generated parts
+```
+
+Opt-in and **billed** (real API calls), so it does _not_ run in CI — the deterministic replay above is the CI gate. Snapshots (hence the judge) need `puppeteer`/Chrome; without them the run scores on auto-verify alone and notes the skipped judge.
+
 ## Programmatic API
 
 ```ts

package/dist/brepjs-verify.cjs

@@ -1,5 +1,5 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_diff = require("./diff-CZ4mLtrf.cjs");
+const require_diff = require("./diff-mxAOOl_m.cjs");
 exports.DEFAULT_TOLERANCE_PCT = require_diff.DEFAULT_TOLERANCE_PCT;
 exports.TYPECHECK_CODE = require_diff.TYPECHECK_CODE;
 exports.emptyReport = require_diff.emptyReport;

package/dist/brepjs-verify.js

@@ -1,2 +1,2 @@
-import { a as typecheckPart, c as isExpectedDims, d as emptyReport, i as TYPECHECK_CODE, l as pctDelta, m as serializeReport, n as runMeasure, o as DEFAULT_TOLERANCE_PCT, r as runPart, s as evaluateExpected, t as runDiff, u as runChecks } from "./diff-D7ZBNRJG.js";
+import { a as typecheckPart, c as isExpectedDims, d as emptyReport, i as TYPECHECK_CODE, l as pctDelta, m as serializeReport, n as runMeasure, o as DEFAULT_TOLERANCE_PCT, r as runPart, s as evaluateExpected, t as runDiff, u as runChecks } from "./diff-Covv1XuJ.js";
 export { DEFAULT_TOLERANCE_PCT, TYPECHECK_CODE, emptyReport, evaluateExpected, isExpectedDims, pctDelta, runChecks, runDiff, runMeasure, runPart, serializeReport, typecheckPart };

package/dist/cli/main.cjs

@@ -1,6 +1,6 @@
 #!/usr/bin/env node
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_diff = require("../diff-CZ4mLtrf.cjs");
+const require_diff = require("../diff-mxAOOl_m.cjs");
 let node_url = require("node:url");
 let node_fs = require("node:fs");
 let node_path = require("node:path");

package/dist/cli/main.js

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { f as pushError, h as loadBrep, m as serializeReport, n as runMeasure, p as reportOk, r as runPart, t as runDiff } from "../diff-D7ZBNRJG.js";
+import { f as pushError, h as loadBrep, m as serializeReport, n as runMeasure, p as reportOk, r as runPart, t as runDiff } from "../diff-Covv1XuJ.js";
 import { fileURLToPath } from "node:url";
 import { existsSync, mkdirSync, realpathSync, watch, writeFileSync } from "node:fs";
 import { basename, dirname, join, resolve } from "node:path";

package/dist/{diff-D7ZBNRJG.js → diff-Covv1XuJ.js}

@@ -39,6 +39,14 @@ function loadBrep() {
 	}
 	return cached;
 }
+var kernelReady;
+function initOcctWasm(brep) {
+	if (!kernelReady) kernelReady = (async () => {
+		const kernel = await (await import("occt-wasm")).OcctKernel.init();
+		brep.registerKernel("occt-wasm", brep.OcctWasmAdapter.fromKernel(kernel));
+	})();
+	return kernelReady;
+}
 //#endregion
 //#region src/verify/report.ts
 function emptyReport() {
@@ -60,7 +68,7 @@ function reportOk(r) {
 	return r.errors.length === 0 && r.checks.every((c) => c.passed) && r.assertions.every((a) => a.passed);
 }
 /**
-* Local, brepjs-cad-owned advice keyed on `BrepErrorCode` values (see `brepjs`'s public
+* Local, brepjs-verify-owned advice keyed on `BrepErrorCode` values (see `brepjs`'s public
 * `BrepErrorCode`). Intentionally not importing the library's internal `getSuggestionForCode`:
 * this table is the agent loop's own actionable `fix` + `nextStep` guidance, and the library's
 * public `BrepError.suggestion` is still surfaced alongside it on each hint.
@@ -68,11 +76,11 @@ function reportOk(r) {
 var HINT_TABLE = {
 	FILLET_NO_EDGES: {
 		fix: "Select real edges before filleting — pass an edge query (e.g. find edges by direction/position) or a non-empty edge list, not the whole solid.",
-		nextStep: "List the solid’s edges, pick the ones to round, then call fillet(solid, radius, edges)."
+		nextStep: "List the solid’s edges, pick the ones to round, then call fillet(solid, edges, radius)."
 	},
 	CHAMFER_NO_EDGES: {
 		fix: "Select real edges before chamfering — pass a non-empty edge query/list rather than relying on a default that matched nothing.",
-		nextStep: "Enumerate the solid’s edges, choose the target edges, then call chamfer(solid, distance, edges)."
+		nextStep: "Enumerate the solid’s edges, choose the target edges, then call chamfer(solid, edges, distance)."
 	},
 	INVALID_FILLET_RADIUS: {
 		fix: "Use a fillet radius that is > 0 and small enough to fit the adjacent faces (well under half the thinnest wall).",
@@ -499,7 +507,7 @@ async function runPart(modulePath, opts = {}) {
 	let brep;
 	try {
 		brep = await loadBrep();
-		await brep.init();
+		await initOcctWasm(brep);
 	} catch (e) {
 		pushError(report, toErrorInfo("kernel init failed", e));
 		return finalize({

package/dist/{diff-CZ4mLtrf.cjs → diff-mxAOOl_m.cjs}

@@ -41,6 +41,14 @@ function loadBrep() {
 	}
 	return cached;
 }
+var kernelReady;
+function initOcctWasm(brep) {
+	if (!kernelReady) kernelReady = (async () => {
+		const kernel = await (await import("occt-wasm")).OcctKernel.init();
+		brep.registerKernel("occt-wasm", brep.OcctWasmAdapter.fromKernel(kernel));
+	})();
+	return kernelReady;
+}
 //#endregion
 //#region src/verify/report.ts
 function emptyReport() {
@@ -62,7 +70,7 @@ function reportOk(r) {
 	return r.errors.length === 0 && r.checks.every((c) => c.passed) && r.assertions.every((a) => a.passed);
 }
 /**
-* Local, brepjs-cad-owned advice keyed on `BrepErrorCode` values (see `brepjs`'s public
+* Local, brepjs-verify-owned advice keyed on `BrepErrorCode` values (see `brepjs`'s public
 * `BrepErrorCode`). Intentionally not importing the library's internal `getSuggestionForCode`:
 * this table is the agent loop's own actionable `fix` + `nextStep` guidance, and the library's
 * public `BrepError.suggestion` is still surfaced alongside it on each hint.
@@ -70,11 +78,11 @@ function reportOk(r) {
 var HINT_TABLE = {
 	FILLET_NO_EDGES: {
 		fix: "Select real edges before filleting — pass an edge query (e.g. find edges by direction/position) or a non-empty edge list, not the whole solid.",
-		nextStep: "List the solid’s edges, pick the ones to round, then call fillet(solid, radius, edges)."
+		nextStep: "List the solid’s edges, pick the ones to round, then call fillet(solid, edges, radius)."
 	},
 	CHAMFER_NO_EDGES: {
 		fix: "Select real edges before chamfering — pass a non-empty edge query/list rather than relying on a default that matched nothing.",
-		nextStep: "Enumerate the solid’s edges, choose the target edges, then call chamfer(solid, distance, edges)."
+		nextStep: "Enumerate the solid’s edges, choose the target edges, then call chamfer(solid, edges, distance)."
 	},
 	INVALID_FILLET_RADIUS: {
 		fix: "Use a fillet radius that is > 0 and small enough to fit the adjacent faces (well under half the thinnest wall).",
@@ -501,7 +509,7 @@ async function runPart(modulePath, opts = {}) {
 	let brep;
 	try {
 		brep = await loadBrep();
-		await brep.init();
+		await initOcctWasm(brep);
 	} catch (e) {
 		pushError(report, toErrorInfo("kernel init failed", e));
 		return finalize({

package/dist/snapshot/registry.cjs

@@ -11,7 +11,7 @@ async function probe(port) {
 		const res = await fetch(`http://127.0.0.1:${port}/__cad/server`, { signal: ctrl.signal });
 		if (!res.ok) return false;
 		const d = await res.json();
-		return d.app === "brepjs-cad-viewer" && d.dynamicRoot === true && typeof d.serverApiVersion === "number" && d.serverApiVersion >= 1;
+		return d.app === "brepjs-verify-viewer" && d.dynamicRoot === true && typeof d.serverApiVersion === "number" && d.serverApiVersion >= 1;
 	} catch {
 		return false;
 	} finally {

package/dist/snapshot/registry.js

@@ -10,7 +10,7 @@ async function probe(port) {
 		const res = await fetch(`http://127.0.0.1:${port}/__cad/server`, { signal: ctrl.signal });
 		if (!res.ok) return false;
 		const d = await res.json();
-		return d.app === "brepjs-cad-viewer" && d.dynamicRoot === true && typeof d.serverApiVersion === "number" && d.serverApiVersion >= 1;
+		return d.app === "brepjs-verify-viewer" && d.dynamicRoot === true && typeof d.serverApiVersion === "number" && d.serverApiVersion >= 1;
 	} catch {
 		return false;
 	} finally {

package/dist/snapshot/static.cjs

@@ -45,7 +45,7 @@ async function handle(req, res, port) {
 	const url = new URL(req.url ?? "/", `http://127.0.0.1:${port}`);
 	if (url.pathname === "/__cad/server") {
 		const d = {
-			app: "brepjs-cad-viewer",
+			app: "brepjs-verify-viewer",
 			port,
 			dynamicRoot: true,
 			serverApiVersion: 1

package/dist/snapshot/static.d.ts

@@ -8,7 +8,7 @@ export interface StaticServer {
     close(): Promise<void>;
 }
 export interface ServerDescriptor {
-    app: 'brepjs-cad-viewer';
+    app: 'brepjs-verify-viewer';
     port: number;
     dynamicRoot: true;
     serverApiVersion: number;

package/dist/snapshot/static.js

@@ -44,7 +44,7 @@ async function handle(req, res, port) {
 	const url = new URL(req.url ?? "/", `http://127.0.0.1:${port}`);
 	if (url.pathname === "/__cad/server") {
 		const d = {
-			app: "brepjs-cad-viewer",
+			app: "brepjs-verify-viewer",
 			port,
 			dynamicRoot: true,
 			serverApiVersion: 1

package/dist/verify/brepjsRuntime.d.ts

@@ -2,3 +2,4 @@ import type * as Brep from 'brepjs';
 export type BrepNs = typeof Brep;
 export declare function toolDir(): string;
 export declare function loadBrep(): Promise<BrepNs>;
+export declare function initOcctWasm(brep: BrepNs): Promise<void>;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "brepjs-verify",
-  "version": "0.2.1",
+  "version": "0.3.0",
   "description": "Agent skill + verify/preview tooling for authoring parametric brepjs CAD code",
   "license": "Apache-2.0",
   "type": "module",
@@ -44,6 +44,7 @@
     "lint": "eslint src tests viewer bench",
     "test": "vitest run",
     "eval": "tsx bench/run.ts",
+    "eval:live": "tsx bench/live.ts",
     "smoke": "node dist/cli/main.js verify tests/fixtures/validBox.brep.ts | node -e \"const r=JSON.parse(require('fs').readFileSync(0,'utf8'));if(r.ok!==true||!(r.measurements.volume>0))process.exit(1)\"",
     "smoke:standalone": "node scripts/smokeStandalone.mjs",
     "prepack": "npm run build"
@@ -58,6 +59,7 @@
     "puppeteer": "^25.0.4"
   },
   "devDependencies": {
+    "@anthropic-ai/sdk": "0.100.1",
     "@react-three/drei": "^10.7.7",
     "@react-three/fiber": "^9.6.1",
     "@types/node": "^25.9.1",
@@ -73,6 +75,7 @@
     "tsx": "^4.22.3",
     "vite": "^8.0.0",
     "vite-plugin-dts": "^5.0.1",
-    "vitest": "^4.0.0"
+    "vitest": "^4.0.0",
+    "zod": "4.4.3"
   }
 }