feat(js/testing): mockModel + echoModel for genkit/testing

[cabljac]

Implementation of the JS testing-module RFC (#5438).

What

Promotes Genkit internal model-mocking helpers into the public genkit/testing surface so app developers can unit-test flows, prompts, tools, and chat deterministically, with no live model, network, or API key.

• mockModel(ai, { respond }) - programmable mock. respond(req, { sendChunk }) drives each call response, including text, structured output, tool requests, or a stream. Typed inspection includes model.lastMessage, model.lastRequest, model.requests, and model.requestCount.
• echoModel(ai) - zero-config model that echoes the rendered request as text, for asserting prompt/message assembly.
• Re-exported from @genkit-ai/ai/testing and genkit/testing, alongside the existing testModels.

Inspection reads in Genkit idiom: model.lastMessage!.text instead of reaching into raw request message content.

Files

File	What
js/ai/src/testing/mock-model.ts	implementation + types
js/ai/src/testing/index.ts, js/genkit/src/testing.ts	re-exports
js/genkit/tests/mock-model_test.ts	unit tests for text, streaming, tool round-trip, inspection, and echo
js/testapps/testing-sample/	sample app with a recommendDish flow and dailySpecial tool
js/testapps/testing-sample/tests/menu_test.ts	Node node:test coverage
js/testapps/testing-sample/tests/menu.vitest.test.ts	vitest coverage

Why a dedicated fake model, not middleware

The RFC alternatives section weighs model middleware as the closest mechanism. This PR implements mockModel and echoModel as real defineModel actions using apiVersion v2, mirroring the existing internal defineProgrammableModel approach.

• Avoids Function.length streaming behavior in model middleware dispatch.
• Keeps typed inspection on the returned model action.
• Promotes existing fake-model logic rather than rebuilding it on a different substrate.

Scope notes

• autoModel is not included; it remains an RFC open question for follow-up.
• The global no-real-model-calls guard is not included in this PR.

Refs #5438.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[gemini-code-assist]

Code Review

This pull request introduces deterministic testing utilities for Genkit, specifically mockModel and echoModel, along with a sample application demonstrating their usage with both node:test and vitest. The review feedback highlights several key robustness and usability improvements in mockModel: handling potential DataCloneErrors when cloning requests with non-serializable properties, joining multi-turn messages with newlines in renderRequestText for better readability, and adding defensive checks in toResponseData to prevent runtime errors when a mock response is falsy.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

Using structuredClone directly on the request object can throw a DataCloneError if the request contains non-serializable properties. In Genkit, the context or config objects often carry complex, non-serializable values (such as database connections, Express request/response objects, or custom class instances).

To make this robust, we should wrap the cloning in a helper that falls back to a safe shallow/semi-deep clone of the messages and parts if structuredClone fails.

    async (request, { sendChunk }) => {
      // Snapshot so later mutation of the request can't alter recorded history.
      let cloned: GenerateRequest;
      try {
        cloned = structuredClone(request);
      } catch {
        cloned = {
          ...request,
          messages: request.messages.map((m) => ({
            ...m,
            content: m.content.map((c) => ({ ...c })),
          })),
        };
      }
      requests.push(cloned);

[gemini-code-assist]

Currently, renderRequestText joins all messages with an empty string (.join('')). This causes multi-turn conversations to run together (e.g., user: hello and model: hi becomes hellohi), making assertions and debugging output hard to read.

Joining the messages with a newline (\n) instead makes the flattened request text much more readable and easier to assert on.

function renderRequestText(request: GenerateRequest): string {
  return request.messages
    .map(
      (m) =>
        (m.role === 'user' || m.role === 'model' ? '' : \`\${m.role}: \`) +
        m.content.map(renderPart).join('')
    )
    .join('\n');
}

[gemini-code-assist]

If the user-defined respond function returns undefined or null (for example, if they only stream chunks or have a void callback), toResponseData will throw a TypeError: Cannot use 'in' operator to search for 'message' in undefined.

We should add a defensive check to handle falsy responses gracefully by returning an empty model response.

function toResponseData(response: MockResponse): GenerateResponseData {
  if (!response) {
    return {
      message: { role: 'model', content: [] },
      finishReason: 'stop',
    };
  }
  if (typeof response === 'string') {
    return {
      message: { role: 'model', content: [{ text: response }] },
      finishReason: 'stop',
    };
  }
  if (typeof response === 'object' && 'message' in response && response.message) {
    return { finishReason: 'stop', ...(response as GenerateResponseData) };
  }