ai 7.0.66 → 7.0.68

package/CHANGELOG.md

@@ -1,5 +1,19 @@
 # ai
 
+## 7.0.68
+
+### Patch Changes
+
+- Updated dependencies [257632b]
+  - @ai-sdk/gateway@4.0.54
+
+## 7.0.67
+
+### Patch Changes
+
+- Updated dependencies [a0b1ffc]
+  - @ai-sdk/gateway@4.0.53
+
 ## 7.0.66
 
 ### Patch Changes

package/dist/index.js

@@ -1143,7 +1143,7 @@ import {
 } from "@ai-sdk/provider-utils";
 
 // src/version.ts
-var VERSION = true ? "7.0.66" : "0.0.0-test";
+var VERSION = true ? "7.0.68" : "0.0.0-test";
 
 // src/util/download/download.ts
 var download = async ({

package/dist/internal/index.js

@@ -92,7 +92,7 @@ import {
 } from "@ai-sdk/provider-utils";
 
 // src/version.ts
-var VERSION = true ? "7.0.66" : "0.0.0-test";
+var VERSION = true ? "7.0.68" : "0.0.0-test";
 
 // src/util/download/download.ts
 var download = async ({

package/docs/02-foundations/02-providers-and-models.mdx

@@ -97,6 +97,7 @@ The open-source community has created the following providers:
 - [ZeroEntropy Provider](/providers/community-providers/zeroentropy) (`zeroentropy-ai-provider`)
 - [Crusoe Provider](/providers/community-providers/crusoe) (`crusoe-ai-provider`)
 - [Neon AI Gateway Provider](/providers/community-providers/neon-ai-gateway) (`@neon/ai-sdk-provider`)
+- [Interfaze Provider](/providers/community-providers/interfaze) (`@interfaze-ai/ai-sdk`)
 
 ## Self-Hosted Models
 

package/docs/03-ai-sdk-core/16-mcp-tools.mdx

@@ -25,6 +25,12 @@ Create an MCP client using one of the following transport options:
 - SSE (Server-Sent Events): An alternative HTTP-based transport
 - `stdio`: For local development only. Uses standard input/output streams for local MCP servers
 
+The AI SDK MCP client supports both legacy initialization-based protocol
+versions and stateless MCP `2026-07-28`. The built-in stdio transport probes
+with `server/discover` and automatically falls back to the legacy
+`initialize` handshake for older servers. Custom transports can opt into this
+negotiation with `supportsProtocolVersionDiscovery: true`.
+
 ### HTTP Transport (Recommended)
 
 For production deployments, we recommend using the HTTP transport. You can configure it directly on the client:
@@ -49,8 +55,9 @@ const mcpClient = await createMCPClient({
 });
 ```
 
-If the MCP server uses Streamable HTTP sessions, you can reattach to a saved
-session by restoring both the previous session id and initialize result:
+If a legacy MCP server uses Streamable HTTP sessions, you can reattach to a
+saved session by restoring both the previous session id and initialize result.
+MCP `2026-07-28` is stateless and does not use these options:
 
 ```typescript
 import { createMCPClient } from '@ai-sdk/mcp';
@@ -203,6 +210,32 @@ This hook is called before the SDK fetches authorization server metadata, so a
 rejected URL is not requested. It is optional and does not change existing OAuth
 flows unless you implement it.
 
+### OAuth Callback Issuer Validation
+
+When completing an OAuth authorization callback, pass the callback's `iss`
+parameter to `auth` as `callbackIssuer`:
+
+```typescript
+import { auth } from '@ai-sdk/mcp';
+
+const callbackUrl = new URL(request.url);
+
+await auth(myOAuthClientProvider, {
+  serverUrl: 'https://mcp.example.com',
+  authorizationCode: callbackUrl.searchParams.get('code')!,
+  callbackState: callbackUrl.searchParams.get('state') ?? undefined,
+  callbackIssuer: callbackUrl.searchParams.get('iss') ?? undefined,
+});
+```
+
+When `iss` is present, it must exactly match the discovered authorization
+server issuer. The authorization code is not exchanged when it does not match.
+
+Dynamic client registration requests include an OAuth `application_type`. The
+client infers `native` for loopback, localhost, and custom-scheme redirects,
+and `web` for remote HTTP(S) redirects. You can override this by setting
+`application_type` in the provider's `clientMetadata`.
+
 ### Retrying Transient Tool Failures
 
 MCP tool calls can fail for transient transport reasons, such as rate limits,

package/docs/03-ai-sdk-core/18-code-mode.mdx

@@ -180,6 +180,17 @@ const result = await runCodeMode({
 `runCodeMode` returns the value returned by the program. It uses the same
 sandbox and execution limits as the AI SDK tool.
 
+## Tool Approval
+
+Code mode does not currently integrate with AI SDK tool approval flows. Tool
+calls made by generated code are nested inside the code mode invocation, so
+they cannot pause the generation and surface a tool approval request to your
+application.
+
+Do not expose tools that rely on user approval to code mode. Keep those tools
+directly callable by the model instead. If a nested tool requires approval,
+the call is rejected rather than executed.
+
 ## Execution Limits
 
 Every invocation has limits for runtime, memory, source size, results, tool
@@ -244,6 +255,3 @@ to code mode.
 
 Tool input schemas are validated before their `execute` functions run. Abort
 signals and AI SDK tool execution context are forwarded to nested tool calls.
-
-Code mode does not currently support approval flows for nested tool calls.
-Tools that require approval are rejected instead of being executed.

package/docs/03-ai-sdk-core/38-video-generation.mdx

@@ -466,21 +466,22 @@ try {
 
 ## Video Models
 
-| Provider                                                                | Model                       | Features                                                                      |
-| ----------------------------------------------------------------------- | --------------------------- | ----------------------------------------------------------------------------- |
-| [FAL](/providers/ai-sdk-providers/fal#video-models)                     | `luma-dream-machine/ray-2`  | Text-to-video, image-to-video                                                 |
-| [FAL](/providers/ai-sdk-providers/fal#video-models)                     | `minimax-video`             | Text-to-video                                                                 |
-| [Google](/providers/ai-sdk-providers/google#video-models)               | `veo-2.0-generate-001`      | Text-to-video, up to 4 videos per call                                        |
-| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models) | `veo-3.1-generate-001`      | Text-to-video, audio generation                                               |
-| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models) | `veo-3.1-fast-generate-001` | Text-to-video, audio generation                                               |
-| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models) | `veo-3.0-generate-001`      | Text-to-video, audio generation                                               |
-| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models) | `veo-3.0-fast-generate-001` | Text-to-video, audio generation                                               |
-| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models) | `veo-2.0-generate-001`      | Text-to-video, up to 4 videos per call                                        |
-| [Kling AI](/providers/ai-sdk-providers/klingai#video-models)            | `kling-v2.6-t2v`            | Text-to-video                                                                 |
-| [Kling AI](/providers/ai-sdk-providers/klingai#video-models)            | `kling-v2.6-i2v`            | Image-to-video                                                                |
-| [Kling AI](/providers/ai-sdk-providers/klingai#video-models)            | `kling-v2.6-motion-control` | Motion control                                                                |
-| [Replicate](/providers/ai-sdk-providers/replicate#video-models)         | `minimax/video-01`          | Text-to-video                                                                 |
-| [xAI](/providers/ai-sdk-providers/xai#video-models)                     | `grok-imagine-video`        | Text-to-video, image-to-video, editing, extension, R2V                        |
-| [xAI](/providers/ai-sdk-providers/xai#video-models)                     | `grok-imagine-video-1.5`    | Text-to-video, image-to-video, editing, extension, R2V (with reference audio) |
+| Provider                                                                        | Model                       | Features                                                                       |
+| ------------------------------------------------------------------------------- | --------------------------- | ------------------------------------------------------------------------------ |
+| [Black Forest Labs](/providers/ai-sdk-providers/black-forest-labs#video-models) | `flux-3-video`              | Text-to-video, image-to-video, keyframes, video continuation, audio generation |
+| [FAL](/providers/ai-sdk-providers/fal#video-models)                             | `luma-dream-machine/ray-2`  | Text-to-video, image-to-video                                                  |
+| [FAL](/providers/ai-sdk-providers/fal#video-models)                             | `minimax-video`             | Text-to-video                                                                  |
+| [Google](/providers/ai-sdk-providers/google#video-models)                       | `veo-2.0-generate-001`      | Text-to-video, up to 4 videos per call                                         |
+| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models)         | `veo-3.1-generate-001`      | Text-to-video, audio generation                                                |
+| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models)         | `veo-3.1-fast-generate-001` | Text-to-video, audio generation                                                |
+| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models)         | `veo-3.0-generate-001`      | Text-to-video, audio generation                                                |
+| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models)         | `veo-3.0-fast-generate-001` | Text-to-video, audio generation                                                |
+| [Google Vertex](/providers/ai-sdk-providers/google-vertex#video-models)         | `veo-2.0-generate-001`      | Text-to-video, up to 4 videos per call                                         |
+| [Kling AI](/providers/ai-sdk-providers/klingai#video-models)                    | `kling-v2.6-t2v`            | Text-to-video                                                                  |
+| [Kling AI](/providers/ai-sdk-providers/klingai#video-models)                    | `kling-v2.6-i2v`            | Image-to-video                                                                 |
+| [Kling AI](/providers/ai-sdk-providers/klingai#video-models)                    | `kling-v2.6-motion-control` | Motion control                                                                 |
+| [Replicate](/providers/ai-sdk-providers/replicate#video-models)                 | `minimax/video-01`          | Text-to-video                                                                  |
+| [xAI](/providers/ai-sdk-providers/xai#video-models)                             | `grok-imagine-video`        | Text-to-video, image-to-video, editing, extension, R2V                         |
+| [xAI](/providers/ai-sdk-providers/xai#video-models)                             | `grok-imagine-video-1.5`    | Text-to-video, image-to-video, editing, extension, R2V (with reference audio)  |
 
 Above are a small subset of the video models supported by the AI SDK providers. For more, see the respective provider documentation.

package/docs/03-ai-sdk-harnesses/02-harness-agent.mdx

@@ -93,6 +93,60 @@ try {
 }
 ```
 
+## Generate Structured Output
+
+Set `output` when constructing `HarnessAgent` to require the same typed output
+on every turn. The agent converts the output specification to JSON Schema for
+the harness adapter, validates the completed response, and returns the parsed
+value through `result.output`.
+
+```ts
+import { HarnessAgent } from '@ai-sdk/harness/agent';
+import { Output } from 'ai';
+import { z } from 'zod';
+
+const agent = new HarnessAgent({
+  harness: claudeCode,
+  sandbox,
+  output: Output.object({
+    schema: z.object({
+      recipe: z.object({
+        name: z.string(),
+        ingredients: z.array(
+          z.object({
+            name: z.string(),
+            amount: z.number(),
+            unit: z.enum(['oz', 'fl oz', 'cup', 'gallon']),
+          }),
+        ),
+        steps: z.array(z.string()),
+      }),
+    }),
+  }),
+});
+
+const session = await agent.createSession();
+try {
+  const result = await agent.generate({
+    session,
+    prompt: 'Generate a lasagna recipe.',
+  });
+  console.dir(result.output, { depth: Infinity });
+} finally {
+  await session.destroy();
+}
+```
+
+With `stream()`, read `partialOutputStream` for incrementally parsed values and
+await `result.output` for the validated final value. Structured data also remains
+available as JSON in the normal text and stream surfaces; adapters do not add it
+to the `finish` part.
+
+Harness structured output requires a schema. Schema-less `Output.json()` and
+adapters or runtime configurations that cannot enforce the schema throw
+`HarnessCapabilityUnsupportedError`; see the
+[adapter capability table](/docs/ai-sdk-harnesses/harness-adapters#adapter-capabilities).
+
 ## Messages and History
 
 A harness session owns its native conversation history. When you pass `messages`
@@ -366,6 +420,7 @@ console.log(preparation.identity);
 - `id`: optional stable agent identifier.
 - `instructions`: instructions appended to the runtime's system or developer
   prompt when supported, or prepended to the first user prompt otherwise.
+- `output`: typed output specification applied to every turn.
 - `stopWhen`: condition(s) for finishing a result slice after a completed
   harness tool step that can continue into another model step.
 - `tools`: AI SDK tools executed by the host when the harness calls them.

package/docs/03-ai-sdk-harnesses/05-harness-adapters.mdx

@@ -30,12 +30,16 @@ The AI SDK includes the following harness adapters:
 
 ## Adapter Capabilities
 
-| Adapter                                                | Runtime location | Custom tools | Custom skills | Built-in tool approval | Built-in tool filtering      |
-| ------------------------------------------------------ | ---------------- | ------------ | ------------- | ---------------------- | ---------------------------- |
-| [Claude Code](/providers/ai-sdk-harnesses/claude-code) | Sandbox bridge   | <Check />    | <Check />     | <Check />              | <Check />                    |
-| [Cline](/providers/ai-sdk-harnesses/cline)             | Host process     | <Check />    | <Check />     | <Check />              | <Check />                    |
-| [Codex](/providers/ai-sdk-harnesses/codex)             | Sandbox bridge   | <Check />    | <Check />     | <Cross />              | <Cross />                    |
-| [Deep Agents](/providers/ai-sdk-harnesses/deepagents)  | Sandbox bridge   | <Check />    | <Check />     | <Check />              | <Check /> via auto-rejection |
-| [Grok Build](/providers/ai-sdk-harnesses/grok-build)   | Sandbox via ACP  | <Check />    | <Check />     | <Check />              | <Cross />                    |
-| [OpenCode](/providers/ai-sdk-harnesses/opencode)       | Sandbox bridge   | <Check />    | <Check />     | <Check />              | <Check /> via auto-rejection |
-| [Pi](/providers/ai-sdk-harnesses/pi)                   | Host process     | <Check />    | <Check />     | <Check />              | <Check />                    |
+| Adapter                                                | Runtime location | Custom tools | Custom skills | Structured output | Built-in tool approval | Built-in tool filtering      |
+| ------------------------------------------------------ | ---------------- | ------------ | ------------- | ----------------- | ---------------------- | ---------------------------- |
+| [Claude Code](/providers/ai-sdk-harnesses/claude-code) | Sandbox bridge   | <Check />    | <Check />     | <Check />         | <Check />              | <Check />                    |
+| [Cline](/providers/ai-sdk-harnesses/cline)             | Host process     | <Check />    | <Check />     | <Check />         | <Check />              | <Check />                    |
+| [Codex](/providers/ai-sdk-harnesses/codex)             | Sandbox bridge   | <Check />    | <Check />     | <Check />         | <Cross />              | <Cross />                    |
+| [Deep Agents](/providers/ai-sdk-harnesses/deepagents)  | Sandbox bridge   | <Check />    | <Check />     | <Check />         | <Check />              | <Check /> via auto-rejection |
+| [Grok Build](/providers/ai-sdk-harnesses/grok-build)   | Sandbox via ACP  | <Check />    | <Check />     | <Check />         | <Check />              | <Cross />                    |
+| [OpenCode](/providers/ai-sdk-harnesses/opencode)       | Sandbox bridge   | <Check />    | <Check />     | <Check />         | <Check />              | <Check /> via auto-rejection |
+| [Pi](/providers/ai-sdk-harnesses/pi)                   | Host process     | <Check />    | <Check />     | <Cross />         | <Check />              | <Check />                    |
+
+Generic ACP profiles do not support structured output unless they configure an
+implementation-specific prompt metadata mapping; Grok Build includes the
+required mapping, while Codex ACP does not expose one.

package/docs/07-reference/01-ai-sdk-core/23-create-mcp-client.mdx

@@ -121,14 +121,14 @@ It currently does not support accepting notifications from an MCP server, and cu
                       type: 'string',
                       isOptional: true,
                       description:
-                        'Initial MCP session id to send with resumed Streamable HTTP requests after initialization. Pair with initialInitializeResult when using createMCPClient. Only used by the HTTP transport.',
+                        'Initial legacy MCP session id to send with resumed Streamable HTTP requests after initialization. Pair with initialInitializeResult when using createMCPClient. Only used by the HTTP transport and initialization-based protocol versions.',
                     },
                     {
                       name: 'initialProtocolVersion',
                       type: 'string',
                       isOptional: true,
                       description:
-                        'Initial MCP protocol version to send before initialize negotiates one. Only used by the HTTP transport.',
+                        'Initial legacy MCP protocol version to send before initialize negotiates one. Only used by the HTTP transport.',
                     },
                     {
                       name: 'onSessionIdChange',
@@ -167,7 +167,7 @@ It currently does not support accepting notifications from an MCP server, and cu
               type: 'RequestOptions',
               isOptional: true,
               description:
-                'Optional signal and timeout settings that bound transport startup and the initialize request. A timeout or abort closes the transport and rejects createMCPClient.',
+                'Optional signal and timeout settings that bound transport startup and protocol negotiation. A timeout or abort closes the transport and rejects createMCPClient.',
             },
             {
               name: 'clientName',
@@ -206,14 +206,14 @@ It currently does not support accepting notifications from an MCP server, and cu
               type: 'InitializeResult',
               isOptional: true,
               description:
-                'Initialize result from a previous MCP session. When provided, the client starts the transport and reuses this metadata without sending a new initialize request.',
+                'Initialize result from a previous legacy MCP session. When provided, the client starts the transport and reuses this metadata without sending a new initialize request.',
             },
             {
               name: 'capabilities',
               type: 'ClientCapabilities',
               isOptional: true,
               description:
-                'Optional client capabilities to advertise during initialization. For example, set { elicitation: {} } to enable handling elicitation requests from the server.',
+                'Optional client capabilities to advertise during legacy initialization or on each stateless modern request. For example, set { elicitation: {} } to enable handling elicitation requests from the server.',
             },
           ],
         },
@@ -232,13 +232,13 @@ Returns a Promise that resolves to an `MCPClient` with the following properties
       name: 'initializeResult',
       type: 'InitializeResult',
       description:
-        'The full initialize result used by this client, either from the server during initialization or from initialInitializeResult.',
+        'Connection metadata used by this client. For legacy servers this is the initialize result; for modern servers an equivalent value is derived from protocol discovery.',
     },
     {
       name: 'serverInfo',
       type: 'Configuration',
       description:
-        'Information about the connected MCP server (name, version, optional title), as reported during the initialize handshake.',
+        'Information about the connected MCP server (name, version, optional title), as reported during initialization or protocol discovery.',
     },
     {
       name: 'instructions',

package/docs/07-reference/04-ai-sdk-workflow/01-workflow-agent.mdx

@@ -404,7 +404,8 @@ To see `WorkflowAgent` in action, check out [these examples](#examples).
       name: 'maxRetries',
       type: 'number',
       isOptional: true,
-      description: 'How many times to retry on failure. Default: 2.',
+      description:
+        'How many times to retry retryable model-call failures. Set to 0 to disable retries. Retry-After response headers are respected, and durable workflow step retries are not stacked. Default: 2.',
     },
     {
       name: 'headers',
@@ -678,6 +679,12 @@ Returns a `Promise<WorkflowAgentStreamResult>` with the following properties:
       description:
         'Tool results from the last step. Only includes results for tools that were executed.',
     },
+    {
+      name: 'error',
+      type: 'unknown | undefined',
+      description:
+        "The original value from a model stream error part. The property is present when an error part was emitted, even if its value is undefined; use `'error' in result` to distinguish that case.",
+    },
     {
       name: 'output',
       type: 'OUTPUT',

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai",
-  "version": "7.0.66",
+  "version": "7.0.68",
   "type": "module",
   "description": "AI SDK by Vercel - build apps like ChatGPT, Claude, Gemini, and more with a single interface for any model using the Vercel AI Gateway or go direct to OpenAI, Anthropic, Google, or any other model provider.",
   "license": "Apache-2.0",
@@ -42,7 +42,7 @@
     }
   },
   "dependencies": {
-    "@ai-sdk/gateway": "4.0.52",
+    "@ai-sdk/gateway": "4.0.54",
     "@ai-sdk/provider": "4.0.7",
     "@ai-sdk/provider-utils": "5.0.27"
   },