ai-retry 1.0.2 → 1.1.0

package/README.md

@@ -358,6 +358,7 @@ There are several built-in dynamic retryables available for common use cases:
 - [`serviceOverloaded`](./src/retryables/service-overloaded.ts): Response with status code 529 (service overloaded).
   - Use this retryable to handle Anthropic's overloaded errors.
 - [`serviceUnavailable`](./src/retryables/service-unavailable.ts): Response with status code 503 (service unavailable).
+- [`schemaMismatch`](./src/retryables/schema-mismatch.ts): Response JSON doesn't match the expected schema from structured output modes (`Output.object()`, `Output.array()`, `Output.choice()`).
 
 #### Content Filter
 
@@ -498,6 +499,46 @@ const retryableModel = createRetryable({
 
 By default, if a [`retry-after-ms`](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/provisioned-get-started#what-should--i-do-when-i-receive-a-429-response) or [`retry-after`](https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Retry-After) header is present in the response, it will be prioritized over the configured delay. The delay from the header will be capped at 60 seconds for safety.
 
+#### Schema Mismatch
+
+Automatically retry with a different model when the response JSON doesn't match the expected schema.
+
+This is a result-based retryable that validates the model's JSON output against the schema set by structured output modes like `Output.object()`, `Output.array()`, and `Output.choice()`. 
+Normally, schema validation happens outside the model in `generateText`, so a schema validation error would not be seen by the retryable model. This retryable catches it early and retries with a fallback model.
+
+> [!NOTE]
+> This retryable works with `generateText` and any structured output mode that provides a schema: `Output.object()`, `Output.array()`, and `Output.choice()`.
+
+```typescript
+import { openai } from '@ai-sdk/openai';
+import { anthropic } from '@ai-sdk/anthropic';
+import { generateText, Output } from 'ai';
+import { createRetryable } from 'ai-retry';
+import { schemaMismatch } from 'ai-retry/retryables';
+import { z } from 'zod';
+
+const retryableModel = createRetryable({
+  model: openai('gpt-4-mini'), // Weak base model
+  retries: [
+    // Retry with stronger model on schema mismatch
+    schemaMismatch(openai('gpt-5')),
+  ],
+});
+
+const result = await generateText({
+  model: retryableModel,
+  output: Output.object({
+    schema: z.object({
+      name: z.string(),
+      age: z.number(),
+    }),
+  }),
+  prompt: 'Generate a person with name and age.',
+});
+
+console.log(result.object); // { name: "Alice", age: 30 }
+```
+
 ### Options
 
 #### Disabling Retries

package/dist/index.d.mts

@@ -1,4 +1,4 @@
-import { C as RetryableModelOptions, S as Retryable, _ as Retry, a as GatewayLanguageModelId, b as RetryErrorAttempt, c as LanguageModelGenerate, d as LanguageModelStreamPart, f as ProviderOptions, g as Retries, h as ResolvedModel, i as EmbeddingModelRetryCallOptions, l as LanguageModelRetryCallOptions, m as ResolvableModel, n as EmbeddingModelCallOptions, o as LanguageModel, p as ResolvableLanguageModel, r as EmbeddingModelEmbed, s as LanguageModelCallOptions, t as EmbeddingModel, u as LanguageModelStream, v as RetryAttempt, w as RetryableOptions, x as RetryResultAttempt, y as RetryContext } from "./types-JvcFQz93.mjs";
+import { C as RetryableModelOptions, S as Retryable, _ as Retry, a as GatewayLanguageModelId, b as RetryErrorAttempt, c as LanguageModelGenerate, d as LanguageModelStreamPart, f as ProviderOptions, g as Retries, h as ResolvedModel, i as EmbeddingModelRetryCallOptions, l as LanguageModelRetryCallOptions, m as ResolvableModel, n as EmbeddingModelCallOptions, o as LanguageModel, p as ResolvableLanguageModel, r as EmbeddingModelEmbed, s as LanguageModelCallOptions, t as EmbeddingModel, u as LanguageModelStream, v as RetryAttempt, w as RetryableOptions, x as RetryResultAttempt, y as RetryContext } from "./types-Bty5BU37.mjs";
 import * as _ai_sdk_provider0 from "@ai-sdk/provider";
 
 //#region src/create-retryable-model.d.ts

package/dist/index.mjs

@@ -1,4 +1,4 @@
-import { a as isLanguageModel, c as isResultAttempt, d as isString, f as isTimeoutError, i as isGenerateResult, l as isStreamContentPart, n as isEmbeddingModel, o as isModel, r as isErrorAttempt, s as isObject, t as isAbortError, u as isStreamResult } from "./utils-Vtk_GA9U.mjs";
+import { a as isLanguageModel, c as isResultAttempt, d as isString, f as isTimeoutError, i as isGenerateResult, l as isStreamContentPart, n as isEmbeddingModel, o as isModel, r as isErrorAttempt, s as isObject, t as isAbortError, u as isStreamResult } from "./utils-fYGELi1C.mjs";
 import { RetryError, gateway } from "ai";
 import { delay } from "@ai-sdk/provider-utils";
 import { getErrorMessage } from "@ai-sdk/provider";
@@ -541,8 +541,8 @@ var RetryableLanguageModel = class {
 				* This will create a new stream.
 				*/
 				const retriedResult = await this.withRetry({
-					fn: async (retryCallOptions$1) => {
-						return this.currentModel.doStream(retryCallOptions$1);
+					fn: async (retryCallOptions) => {
+						return this.currentModel.doStream(retryCallOptions);
 					},
 					callOptions,
 					attempts,

package/dist/retryables/index.d.mts

@@ -1,7 +1,6 @@
-import { S as Retryable, p as ResolvableLanguageModel, t as EmbeddingModel, w as RetryableOptions } from "../types-JvcFQz93.mjs";
+import { S as Retryable, p as ResolvableLanguageModel, t as EmbeddingModel, w as RetryableOptions } from "../types-Bty5BU37.mjs";
 
 //#region src/retryables/content-filter-triggered.d.ts
-
 /**
  * Fallback to a different model if the content filter was triggered.
  */
@@ -29,6 +28,28 @@ declare function requestTimeout<MODEL extends ResolvableLanguageModel | Embeddin
  */
 declare function retryAfterDelay<MODEL extends ResolvableLanguageModel | EmbeddingModel>(options: RetryableOptions<MODEL>): Retryable<MODEL>;
 //#endregion
+//#region src/retryables/schema-mismatch.d.ts
+/**
+ * Fallback to a different model if the response does not match the expected JSON schema.
+ *
+ * Validates the response text against the JSON schema from `responseFormat.schema`
+ * (set automatically when using `Output.object()`) and retries with a different model
+ * if validation fails.
+ *
+ * @example
+ * ```ts
+ * const result = await generateText({
+ *   model: createRetryable({
+ *     model: primaryModel,
+ *     retries: [schemaMismatch(fallbackModel)],
+ *   }),
+ *   output: Output.object({ schema: z.object({ name: z.string() }) }),
+ *   prompt: `Generate a person`,
+ * });
+ * ```
+ */
+declare function schemaMismatch<MODEL extends ResolvableLanguageModel>(model: MODEL, options?: RetryableOptions<MODEL>): Retryable<MODEL>;
+//#endregion
 //#region src/retryables/service-overloaded.d.ts
 /**
  * Fallback to a different model if the provider returns an overloaded error.
@@ -46,4 +67,4 @@ declare function serviceOverloaded<MODEL extends ResolvableLanguageModel | Embed
  */
 declare function serviceUnavailable<MODEL extends ResolvableLanguageModel | EmbeddingModel>(model: MODEL, options?: RetryableOptions<MODEL>): Retryable<MODEL>;
 //#endregion
-export { contentFilterTriggered, requestNotRetryable, requestTimeout, retryAfterDelay, serviceOverloaded, serviceUnavailable };
+export { contentFilterTriggered, requestNotRetryable, requestTimeout, retryAfterDelay, schemaMismatch, serviceOverloaded, serviceUnavailable };

package/dist/retryables/index.mjs

@@ -1,5 +1,7 @@
-import { c as isResultAttempt, d as isString, f as isTimeoutError, r as isErrorAttempt, s as isObject } from "../utils-Vtk_GA9U.mjs";
+import { c as isResultAttempt, d as isString, f as isTimeoutError, r as isErrorAttempt, s as isObject } from "../utils-fYGELi1C.mjs";
 import { APICallError } from "ai";
+import { safeParseJSON } from "@ai-sdk/provider-utils";
+import { fromJSONSchema } from "zod";
 
 //#region src/retryables/content-filter-triggered.ts
 /**
@@ -120,6 +122,52 @@ function retryAfterDelay(options) {
 	};
 }
 
+//#endregion
+//#region src/retryables/schema-mismatch.ts
+/**
+* Fallback to a different model if the response does not match the expected JSON schema.
+*
+* Validates the response text against the JSON schema from `responseFormat.schema`
+* (set automatically when using `Output.object()`) and retries with a different model
+* if validation fails.
+*
+* @example
+* ```ts
+* const result = await generateText({
+*   model: createRetryable({
+*     model: primaryModel,
+*     retries: [schemaMismatch(fallbackModel)],
+*   }),
+*   output: Output.object({ schema: z.object({ name: z.string() }) }),
+*   prompt: `Generate a person`,
+* });
+* ```
+*/
+function schemaMismatch(model, options) {
+	return async (context) => {
+		/**
+		* Only handle result attempts
+		*/
+		if (!isResultAttempt(context.current)) return void 0;
+		const { result, options: callOptions } = context.current;
+		/** Extract text from content */
+		const text = result.content.filter((part) => part.type === `text`).map((part) => part.text).join(``);
+		if (!text) return void 0;
+		/** Auto-detect schema from responseFormat (set by Output.object()) */
+		const responseFormat = callOptions.responseFormat;
+		if (responseFormat?.type !== `json` || !responseFormat.schema) return;
+		if ((await safeParseJSON({
+			text,
+			schema: fromJSONSchema(responseFormat.schema)
+		})).success) return void 0;
+		return {
+			model,
+			maxAttempts: 1,
+			...options
+		};
+	};
+}
+
 //#endregion
 //#region src/retryables/service-overloaded.ts
 /**
@@ -171,4 +219,4 @@ function serviceUnavailable(model, options) {
 }
 
 //#endregion
-export { contentFilterTriggered, requestNotRetryable, requestTimeout, retryAfterDelay, serviceOverloaded, serviceUnavailable };
+export { contentFilterTriggered, requestNotRetryable, requestTimeout, retryAfterDelay, schemaMismatch, serviceOverloaded, serviceUnavailable };

package/package.json

@@ -1,9 +1,7 @@
 {
   "name": "ai-retry",
-  "version": "1.0.2",
+  "version": "1.1.0",
   "description": "AI SDK Retry",
-  "main": "./dist/index.mjs",
-  "module": "./dist/index.mjs",
   "types": "./dist/index.d.mts",
   "type": "module",
   "files": [
@@ -33,30 +31,30 @@
     "ai": "6.x"
   },
   "devDependencies": {
-    "@ai-sdk/anthropic": "3.0.9",
-    "@ai-sdk/azure": "3.0.7",
-    "@ai-sdk/gateway": "3.0.10",
-    "@ai-sdk/groq": "3.0.4",
-    "@ai-sdk/openai": "3.0.7",
-    "@ai-sdk/test-server": "1.0.1",
+    "@ai-sdk/anthropic": "3.0.23",
+    "@ai-sdk/azure": "3.0.19",
+    "@ai-sdk/gateway": "3.0.23",
+    "@ai-sdk/groq": "3.0.15",
+    "@ai-sdk/openai": "3.0.19",
+    "@ai-sdk/test-server": "1.0.3",
     "@arethetypeswrong/cli": "^0.18.2",
-    "@biomejs/biome": "^2.3.10",
+    "@biomejs/biome": "^2.3.13",
     "@total-typescript/tsconfig": "^1.0.4",
-    "@types/node": "^25.0.3",
-    "ai": "6.0.20",
+    "@types/node": "^25.0.10",
+    "ai": "6.0.50",
     "husky": "^9.1.7",
-    "msw": "^2.12.4",
+    "msw": "^2.12.7",
     "pkg-pr-new": "^0.0.62",
-    "publint": "^0.3.16",
-    "tsdown": "^0.18.2",
+    "publint": "^0.3.17",
+    "tsdown": "^0.20.1",
     "tsx": "^4.21.0",
     "typescript": "^5.9.3",
-    "vitest": "^4.0.16",
-    "zod": "^4.2.1"
+    "vitest": "^4.0.18"
   },
   "dependencies": {
-    "@ai-sdk/provider": "3.0.2",
-    "@ai-sdk/provider-utils": "4.0.4"
+    "@ai-sdk/provider": "3.0.5",
+    "@ai-sdk/provider-utils": "4.0.9",
+    "zod": "^4.3.6"
   },
   "scripts": {
     "build": "tsdown",