tests/client.test.ts (1)

1638-1641: Make result assertions order-independent.

These assertions assume a stable response array order. Matching by correlationId will make the test less brittle.

♻️ Suggested test update

-        expect(response.result[0].allowed).toBe(true);
-        expect(response.result[1].allowed).toBe(false);
+        const cor1 = response.result.find((r) => r.correlationId === "cor-1");
+        const cor2 = response.result.find((r) => r.correlationId === "cor-2");
+        expect(cor1?.allowed).toBe(true);
+        expect(cor2?.allowed).toBe(false);

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/client.test.ts` around lines 1638 - 1641, The test currently asserts
response.result items by array index which is brittle; update the assertions to
be order-independent by locating items by their correlationId instead of array
positions. In the test block that checks scope.isDone() and response.result,
find the two result objects using response.result.find(r => r.correlationId ===
'<expected-id>') (or similar) and assert their allowed properties (e.g.,
foundA.allowed.toBe(true), foundB.allowed.toBe(false)); keep the existing
scope.isDone() assertion and replace indexed checks on response.result[0] and
response.result[1] with these correlationId-based lookups.

client.ts (1)

732-778: Resolve authorization model ID once before batching.

batchCheck currently resolves/validates the model ID inside each batch callback. Resolving it once up front is simpler and fails fast before scheduling batch work.

♻️ Suggested refactor

   const {
     headers = {},
     maxBatchSize = DEFAULT_MAX_BATCH_SIZE,
     maxParallelRequests = DEFAULT_MAX_METHOD_PARALLEL_REQS,
   } = options;
+  const authorizationModelId = this.getAuthorizationModelId(options);

   setHeaderIfNotSet(headers, CLIENT_BULK_REQUEST_ID_HEADER, generateRandomIdWithNonUniqueFallback());
@@
     const batchRequest: BatchCheckRequest = {
       checks: batch,
-      authorization_model_id: this.getAuthorizationModelId(options),
+      authorization_model_id: authorizationModelId,
       consistency: options.consistency,
     };

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@client.ts` around lines 732 - 778, In batchCheck, call
this.getAuthorizationModelId(options) once before creating batchedChecks and
store the result (e.g., authorizationModelId) so validation/resolution happens
up-front and can fail fast; then use that stored authorizationModelId inside the
batchResponses async callback when building batchRequest instead of calling
this.getAuthorizationModelId(options) per-batch. Ensure the variable is
validated/awaited before asyncPool is invoked and referenced in the batchRequest
object alongside options.consistency.