Back to Copilotkit

Writing agent state

showcase/shell-docs/src/content/docs/integrations/mastra/shared-state/in-app-agent-write.mdx

1.57.04.2 KB
Original Source

<IframeSwitcher id="shared-state-write-example" exampleUrl="https://feature-viewer.copilotkit.ai/mastra/feature/shared_state?sidebar=false&chatDefaultOpen=false" codeUrl="https://feature-viewer.copilotkit.ai/mastra/feature/shared_state?view=code&sidebar=false&codeLayout=tabs" exampleLabel="Demo" codeLabel="Code" height="700px" />

<Callout type="info"> This example demonstrates writing to shared state in the [CopilotKit Feature Viewer](https://feature-viewer.copilotkit.ai/mastra/feature/shared_state). </Callout>

What is this?

You can easily write to your agent's state from your native application, allowing you to update the agent's working memory from your UI.

<Callout type="info" title="Important"> This guide assumes you are embedding your Mastra agent inside of Copilot Runtime, like so.
ts
const runtime = new CopilotRuntime({
  agents: MastraAgent.getLocalAgents({ mastra }),
});

This feature will not work if you are using a remote Mastra agent.

</Callout>

When should I use this?

You can use this when you want to provide user input or control to your agent's working memory. As your application state changes, you can update the agent state to reflect these changes.

Implementation

<Steps> <Step> ### Run and connect your agent <RunAndConnect /> </Step> <Step> ### Define the Agent State Mastra has advanced [working memory concepts](https://mastra.ai/en/docs/memory/working-memory) to provide statefulness to your agents. CopilotKit leverages Mastra's working memory concept to allow you to implement shared state between your agent and your application.
Providing working memory to your agent is as simple as providing a Zod schema to your agent.

```ts title="mastra/agents/language-agent.ts"

// [!code highlight:4]
// 1. Define the agent state schema
export const AgentStateSchema = z.object({
  language: z.enum(["english", "spanish"]),
});

// 2. Infer the agent state type from the schema
export const AgentState = z.infer<typeof AgentStateSchema>;

// 3. Create the agent
export const languageAgent = new Agent({
  name: "Language Agent",
  model: openai("gpt-5.4"),
  instructions: "Always communicate in the preferred language of the user as defined in your working memory. Do not communicate in any other language.",
  memory: new Memory({
    storage: new LibSQLStore({ id: "mastra-storage", url: ":memory:" }),
    options: {
      // [!code highlight:4]
      workingMemory: {
        enabled: true,
        schema: AgentStateSchema,
      },
    },
  }),
});
```
</Step> <Step> ### Use the `useAgent` Hook With your agent connected and running all that is left is to call the `useAgent` hook, pass the agent's name, and use `agent.setState` to update the agent state.
```tsx title="ui/app/page.tsx"

function YourMainContent() {
  // [!code highlight:5]
  const { agent } = useAgent({
    agentId: "your-mastra-agent-name",
    // optionally provide a type-safe initial state
    initialState: { language: "english" }
  });

  const toggleLanguage = () => {
    agent.setState({ language: agent.state?.language === "english" ? "spanish" : "english" }); // [!code highlight]
  };

  return (
    // style excluded for brevity
    <div>
      <h1>Your main content</h1>
      <p>Language: {agent.state?.language}</p>
      <button onClick={toggleLanguage}>Toggle Language</button>
    </div>
  );
}
```
<Callout type="info">
  The `agent.setState` function in `useAgent` will update the working memory and trigger a rerender when the state changes.
</Callout>
</Step> <Step> ### Give it a try! You can now use the `setState` function to update the agent state and `state` to read it. Try toggling the language button and talking to your agent. You'll see the language change to match the agent's state.
<Callout type="info" title="Important">
  Shared-state in Mastra is prompt-driven. This means that your Agent's awareness of the shared state will be augmented
  by the instructions you provide to your agent.
</Callout>
</Step> </Steps>