### Start NodeSDK and Instrument AI SDK Calls

Source: https://docs.thecontext.company/frameworks/ai-sdk/setup

Demonstrates starting the OpenTelemetry NodeSDK with the `TCCSpanProcessor` and then making AI SDK calls with telemetry enabled. Includes examples for basic usage and integration with an Express server.

```typescript
import { generateText } from "ai";
import { tcc } from "./tcc";

// Start listening for traces
tcc.start();

async function main() {
  const result = await generateText({
    // ...
    experimental_telemetry: { isEnabled: true }, // required
  });
}
```

```typescript
import { streamText } from "ai";
import express, { Request, Response } from "express";
import { tcc } from "./tcc";

const app = express();

// Start listening for traces
tcc.start();

app.post("/", async (req: Request, res: Response) => {
  const result = streamText({
    // ...
    experimental_telemetry: { isEnabled: true }, // required
  });
  result.pipeUIMessageStreamToResponse(res);
});

app.listen(8080, () => {
  console.log(`Server listening on port ${8080}`);
});
```

--------------------------------

### Install LangChain Observability Dependencies (poetry)

Source: https://docs.thecontext.company/frameworks/langchain

Installs the necessary Python package for LangChain observability using poetry.

```bash
poetry add "tcc-otel[langchain]"
```

--------------------------------

### Install Dependencies (npm)

Source: https://docs.thecontext.company/frameworks/ai-sdk/local

Installs the necessary @contextcompany/otel, @vercel/otel, and @opentelemetry/api packages using npm.

```bash
npm i @contextcompany/otel @vercel/otel @opentelemetry/api
```

--------------------------------

### Install LangChain Observability Dependencies (pip)

Source: https://docs.thecontext.company/frameworks/langchain

Installs the necessary Python package for LangChain observability using pip.

```bash
pip install "tcc-otel[langchain]"
```

--------------------------------

### Install Mastra Dependencies (pnpm, npm)

Source: https://docs.thecontext.company/frameworks/mastra/setup

Installs the necessary Mastra package for your project using either pnpm or npm. This package provides the core functionality for integrating with The Context Company.

```shell
pnpm add @contextcompany/mastra
```

```shell
npm i @contextcompany/mastra
```

--------------------------------

### Install Claude Agent SDK Dependencies

Source: https://docs.thecontext.company/frameworks/claude-agent-sdk/setup

Install the necessary packages for integrating The Context Company's SDK with the Claude Agent SDK. Supports both pnpm and npm package managers.

```pnpm
pnpm add @contextcompany/claude @anthropic-ai/claude-agent-sdk
```

```npm
npm i @contextcompany/claude @anthropic-ai/claude-agent-sdk
```

--------------------------------

### Install Dependencies (pnpm)

Source: https://docs.thecontext.company/frameworks/ai-sdk/local

Installs the necessary @contextcompany/otel, @vercel/otel, and @opentelemetry/api packages using pnpm.

```bash
pnpm add @contextcompany/otel @vercel/otel @opentelemetry/api
```

--------------------------------

### Install Dependencies for Next.js and Node.js

Source: https://docs.thecontext.company/frameworks/vercel

Installs the necessary packages for instrumenting AI SDK calls with The Context Company. This includes the `@contextcompany/otel` package and OpenTelemetry core packages.

```bash
pnpm add @contextcompany/otel @vercel/otel @opentelemetry/api
```

```bash
npm add @contextcompany/otel @vercel/otel @opentelemetry/api
```

--------------------------------

### Install AI SDK Dependencies (Node.js)

Source: https://docs.thecontext.company/frameworks/ai-sdk/setup

Installs the necessary packages for AI SDK instrumentation in a Node.js application using either pnpm or npm.

```shell
pnpm add @contextcompany/otel @opentelemetry/sdk-node @opentelemetry/api
```

```shell
npm i @contextcompany/otel @opentelemetry/sdk-node @opentelemetry/api
```

--------------------------------

### Install Dependencies for Node.js

Source: https://docs.thecontext.company/frameworks/vercel

Installs the necessary packages for instrumenting AI SDK calls within a Node.js environment. This includes the `@contextcompany/otel` package and OpenTelemetry Node.js SDK components.

```bash
pnpm add @contextcompany/otel @opentelemetry/sdk-node @opentelemetry/api
```

```bash
npm add @contextcompany/otel @opentelemetry/sdk-node @opentelemetry/api
```

--------------------------------

### Install AI SDK Dependencies (Next.js)

Source: https://docs.thecontext.company/frameworks/ai-sdk/setup

Installs the necessary packages for AI SDK instrumentation in a Next.js application using either pnpm or npm.

```shell
pnpm add @contextcompany/otel @vercel/otel @opentelemetry/api
```

```shell
npm i @contextcompany/otel @vercel/otel @opentelemetry/api
```

--------------------------------

### Start OpenTelemetry NodeSDK and Instrument AI SDK Calls

Source: https://docs.thecontext.company/frameworks/vercel

Starts the OpenTelemetry `NodeSDK` with the configured `TCCSpanProcessor` and demonstrates how to instrument AI SDK calls within a Node.js application. This ensures that traces generated by AI SDK calls are captured and sent to The Context Company.

```typescript
import { generateText } from "ai";
import { tcc } from "./tcc";

// Start listening for traces
tcc.start();

async function main() {
  const result = await generateText({
    // ...
    experimental_telemetry: { isEnabled: true }, // required
  });
}
```

--------------------------------

### Instrument Claude Agent SDK with The Context Company

Source: https://docs.thecontext.company/frameworks/claude-agent-sdk/setup

Wrap the Claude Agent SDK with the `instrumentClaudeAgent` function from The Context Company's SDK. This enables automatic export of agent-related traces to The Context Company platform.

```typescript
import { instrumentClaudeAgent } from "@contextcompany/claude";
import * as claudeSDK from "@anthropic-ai/claude-agent-sdk";

const { query, tool, createSdkMcpServer } = instrumentClaudeAgent(claudeSDK);

// Now use query, tool, and createSdkMcpServer as you normally would
```

--------------------------------

### Install LangGraph Observability Dependencies

Source: https://docs.thecontext.company/frameworks/langgraph

Installs the necessary TCC OpenTelemetry package for LangGraph and LangChain integration. Supports both pip and poetry package managers.

```shell
pip install "tcc-otel[langchain]"
```

```shell
poetry add "tcc-otel[langchain]"
```

--------------------------------

### Submit User Feedback for Agent Runs (TypeScript)

Source: https://docs.thecontext.company/frameworks/mastra/usage

Provides a code example for submitting user feedback using the `submitFeedback` function from `@contextcompany/mastra`. This function takes a `runId`, an optional `score` (thumbs up/down), and optional `text` feedback. At least one of `score` or `text` must be provided. The feedback is associated with the agent run identified by `runId`.

```typescript
import { submitFeedback } from "@contextcompany/mastra";

// Submit both score and text feedback:
await submitFeedback({
  runId: runId, // The run ID from your agent execution
  score: "thumbs_up", // Optional: "thumbs_up" or "thumbs_down"
  text: "This was a helpful response!", // Optional: up to 2000 characters
});
```

--------------------------------

### Instrument Mastra with TCC Exporter (TypeScript)

Source: https://docs.thecontext.company/frameworks/mastra/setup

Integrates The Context Company's Mastra exporter into your Mastra configuration. This allows Mastra agent runs, steps, and tool calls to be exported as traces to The Context Company dashboard.

```typescript
import { Mastra } from "@mastra/core/mastra";
import { exampleAgent } from "./agents/example-agent";
import { TCCMastraExporter } from "@contextcompany/mastra";

// Initialize Mastra with TCC observability exporter
export const mastra = new Mastra({
  agents: { exampleAgent },
  observability: {
    configs: {
      otel: {
        serviceName: "mastra-weather-agent",
        exporters: [new TCCMastraExporter({})],
      },
    },
  },
});
```

--------------------------------

### Set TCC Environment Variables (.env.local)

Source: https://docs.thecontext.company/frameworks/mastra/setup

Sets the TCC_API_KEY environment variable, which is used by The Context Company SDKs to authenticate API requests. Ensure this key is kept confidential.

```env
TCC_API_KEY="your-api-key"
```

--------------------------------

### Instrument LangChain Application

Source: https://docs.thecontext.company/frameworks/langchain

Initializes LangChain instrumentation before importing LangChain. Requires TCC_API_KEY environment variable. This code snippet demonstrates setting up observability for LangChain applications.

```python
from dotenv import load_dotenv
import os
load_dotenv()

# Import and initialize TCC instrumentation BEFORE importing LangChain
from tcc_otel import instrument_langchain

instrument_langchain(
    api_key=os.getenv("TCC_API_KEY"),
)

# Your existing LangChain code
from langchain_openai import ChatOpenAI
from langchain.agents import create_openai_functions_agent, AgentExecutor

model = ChatOpenAI(model="gpt-4")
agent = create_openai_functions_agent(model, tools=[])
executor = AgentExecutor(agent=agent, tools=[])
```

--------------------------------

### Submit User Feedback for Agent Runs (TypeScript)

Source: https://docs.thecontext.company/frameworks/ai-sdk/usage

Illustrates how to submit user feedback, including a score (thumbs up/down) and optional text comments, for a specific agent run using its `runId`. This function requires the '@contextcompany/otel' package and allows for collecting user satisfaction data.

```typescript
import { submitFeedback } from "@contextcompany/otel";

// Submit both score and text feedback:
await submitFeedback({
  runId: runId, // The run ID from your agent execution
  score: "thumbs_up", // Optional: "thumbs_up" or "thumbs_down"
  text: "This was a helpful response!", // Optional: up to 2000 characters
});
```

--------------------------------

### Add Custom Metadata to LangChain Agent Run

Source: https://docs.thecontext.company/frameworks/langchain

Demonstrates how to add custom metadata to LangChain agent runs by passing it through LangChain's RunnableConfig. This allows for filtering and analysis of agent runs based on custom properties.

```python
result = agent_executor.invoke(
    {"input": "Hello!"},
    {
        "metadata": {
            # e.g. tag this agent run with a user id
            "userId": "4a6b111c-b53a-4d00-a877-67185022ab9e",
        }
    }
)
```

--------------------------------

### Add Custom Metadata to Agent Runs (TypeScript)

Source: https://docs.thecontext.company/frameworks/mastra/usage

Demonstrates how to add custom key-value metadata to agent runs using the `tracingOptions.metadata` object. This allows for filtering and analysis of agent runs based on custom business logic, such as user IDs or feature tags. Ensure the metadata is correctly formatted as a key-value pair.

```typescript
await agent.stream(
  // ...
  {
    // Pass metadata to track and filter this execution
    tracingOptions: {
      metadata: {
        // e.g. tag this agent run with a user id
        userId: "4a6b111c-b53a-4d00-a877-67185022ab9e",
        customMetadata1: "1234567890",
        customMetadata2: "context is key",
      },
    },
  }
);
```

--------------------------------

### Enable Debug Mode for OTel in Node.js (TypeScript)

Source: https://docs.thecontext.company/frameworks/ai-sdk/usage

This snippet demonstrates how to enable debug mode for OpenTelemetry tracing in a standard Node.js environment. It involves creating a `TCCSpanProcessor` with `debug: true` and configuring the `NodeSDK`. It requires the `@opentelemetry/sdk-node` and `@contextcompany/otel` packages.

```typescript
import { TCCSpanProcessor } from "@contextcompany/otel";
import { NodeSDK } from "@opentelemetry/sdk-node";

export const tcc = new NodeSDK({
  spanProcessors: [new TCCSpanProcessor({ debug: true })],
});
```

--------------------------------

### Configure NodeSDK with TCCSpanProcessor

Source: https://docs.thecontext.company/frameworks/vercel

Sets up a `NodeSDK` instance for Node.js applications, incorporating the `TCCSpanProcessor`. This configuration is typically placed in a dedicated file (e.g., `tcc.ts`) for easy management and reuse.

```typescript
import { TCCSpanProcessor } from "@contextcompany/otel";
import { NodeSDK } from "@opentelemetry/sdk-node";

export const tcc = new NodeSDK({
  spanProcessors: [new TCCSpanProcessor()],
});
```

--------------------------------

### Add Custom Metadata to Agent Runs (TypeScript)

Source: https://docs.thecontext.company/frameworks/ai-sdk/usage

Demonstrates how to add custom metadata to agent runs using the Vercel AI SDK. This allows for tagging agent runs with specific identifiers like userId for filtering and analysis in the dashboard. It requires the '@ai-sdk/openai' and 'ai' packages.

```typescript
import { openai } from "@ai-sdk/openai";
import { generateText } from "ai";

const { text } = await generateText({
  // ...
  experimental_telemetry: {
    isEnabled: true,
    metadata: {
      // e.g. tag this agent run with a user id
      userId: "4a6b111c-b53a-4d00-a877-67185022ab9e",
    },
  },
});
```

--------------------------------

### Instrument Next.js with TCCSpanProcessor and @vercel/otel

Source: https://docs.thecontext.company/frameworks/vercel

Provides an alternative method for instrumenting Next.js applications by directly using `TCCSpanProcessor` with the `registerOTel` function from `@vercel/otel`. This allows for additional span processors to be configured.

```typescript
import { TCCSpanProcessor } from "@contextcompany/otel";
import { registerOTel } from "@vercel/otel";

export function register() {
  if (process.env.NEXT_RUNTIME === "nodejs")
    registerOTel({
      spanProcessors: [new TCCSpanProcessor()],
    });
}
```

--------------------------------

### Enable Debug Mode for OTel in Next.js (TypeScript)

Source: https://docs.thecontext.company/frameworks/ai-sdk/usage

This code snippet enables debug mode for OpenTelemetry tracing within a Next.js application's Node.js runtime. It uses the `@contextcompany/otel/nextjs` package to register the OTel configuration. Ensure you are within the `process.env.NEXT_RUNTIME === "nodejs"` check.

```typescript
import { registerOTelTCC } from "@contextcompany/otel/nextjs";

export function register() {
  if (process.env.NEXT_RUNTIME === "nodejs") {
    registerOTelTCC({ debug: true });
  }
}
```

--------------------------------

### Generate and Pass Run ID for User Feedback (TypeScript)

Source: https://docs.thecontext.company/frameworks/claude-agent-sdk/usage

Shows how to generate a unique run ID (UUID) and pass it to the `query` function for tracking agent runs that will later receive user feedback. This involves using `randomUUID` from 'crypto' and `instrumentClaudeAgent`.

```typescript
import { instrumentClaudeAgent } from "@contextcompany/claude";
import * as claudeSDK from "@anthropic-ai/claude-agent-sdk";
import { randomUUID } from "crypto";

const { query } = instrumentClaudeAgent(claudeSDK);

// Generate a unique run ID (must be a UUID) before your agent execution
const runId = randomUUID();

const result = query({
  // ...
  tcc: {
    runId: runId, // Pass the run ID
  },
});

// Return the runId to your client
return { result, runId };
```

--------------------------------

### Add Custom Metadata to Agent Runs (TypeScript)

Source: https://docs.thecontext.company/frameworks/claude-agent-sdk/usage

Demonstrates how to add custom key-value metadata to agent runs using the Claude Agent SDK. This allows for filtering and analysis based on custom dimensions like user ID. It requires importing `instrumentClaudeAgent` from '@contextcompany/claude' and the Claude SDK.

```typescript
import { instrumentClaudeAgent } from "@contextcompany/claude";
import * as claudeSDK from "@anthropic-ai/claude-agent-sdk";

const { query } = instrumentClaudeAgent(claudeSDK);

const result = query({
  // ...
  tcc: {
    metadata: {
      // e.g. tag this agent run with a user id
      userId: "4a6b111c-b53a-4d00-a877-67185022ab9e",
    },
  },
});
```

--------------------------------

### Initialize LangGraph Instrumentation

Source: https://docs.thecontext.company/frameworks/langgraph

Initializes TCC OpenTelemetry instrumentation for LangGraph and LangChain. This must be done before importing LangGraph or LangChain. Requires TCC_API_KEY environment variable.

```python
from dotenv import load_dotenv
import os
load_dotenv()

# Import and initialize TCC instrumentation BEFORE importing LangGraph
from tcc_otel import instrument_langchain

instrument_langchain(
    api_key=os.getenv("TCC_API_KEY"),
)

# Your existing LangChain code
from langchain_openai import ChatOpenAI
from langgraph.prebuilt import create_react_agent

model = ChatOpenAI(model="gpt-4")
agent = create_react_agent(model, tools=[])
```

--------------------------------

### Submit User Feedback for Agent Runs (TypeScript)

Source: https://docs.thecontext.company/frameworks/claude-agent-sdk/usage

Illustrates how to submit user feedback, including an optional score ('thumbs_up' or 'thumbs_down') and text, for a specific agent run using the `submitFeedback` function. This requires the `runId` generated previously.

```typescript
import { submitFeedback } from "@contextcompany/claude";

// Submit both score and text feedback:
await submitFeedback({
  runId: runId, // The run ID from your agent execution
  score: "thumbs_up", // Optional: "thumbs_up" or "thumbs_down"
  text: "This was a helpful response!", // Optional: up to 2000 characters
});
```

--------------------------------

### Enable Debug Mode for Mastra Exporter (TypeScript)

Source: https://docs.thecontext.company/frameworks/mastra/usage

Shows how to enable debug mode for the `TCCMastraExporter` by setting the `debug` option to `true` within the observability configuration. This will log all created and exported spans, aiding in debugging and understanding the agent's execution flow. Ensure the Mastra instance is configured with the `TCCMastraExporter`.

```typescript
import { Mastra } from "@mastra/core/mastra";
import { exampleAgent } from "./agents/example-agent";
import { TCCMastraExporter } from "@contextcompany/mastra";

export const mastra = new Mastra({
  agents: { exampleAgent },
  observability: {
    configs: {
      otel: {
        serviceName: "mastra-weather-agent",
        exporters: [new TCCMastraExporter({ debug: true })], // Add debug: true here
      },
    },
  },
});
```

--------------------------------

### Track Agent Sessions with Session ID (TypeScript)

Source: https://docs.thecontext.company/frameworks/claude-agent-sdk/usage

Explains how to group multiple agent runs into a single agent session by providing a `sessionId` within the `tcc` parameter of the `query` function. This is useful for tracking conversations. It uses `instrumentClaudeAgent` from '@contextcompany/claude'.

```typescript
import { instrumentClaudeAgent } from "@contextcompany/claude";
import * as claudeSDK from "@anthropic-ai/claude-agent-sdk";

const { query } = instrumentClaudeAgent(claudeSDK);

const result = query({
  // ...
  tcc: {
    sessionId: "some-session-id", // Track agent sessions
  },
});
```

--------------------------------

### Enable Telemetry for AI SDK Calls (Next.js)

Source: https://docs.thecontext.company/frameworks/ai-sdk/setup

Enables experimental telemetry for AI SDK calls in Next.js by setting the `experimental_telemetry.isEnabled` flag to `true` within `generateText` or `streamText` configurations.

```typescript
import { generateText } from "ai";

const result = generateText({
  // ...
  experimental_telemetry: { isEnabled: true }, // required
});
```

```typescript
import { streamText } from "ai";

const result = streamText({
  // ...
  experimental_telemetry: { isEnabled: true }, // required
});
```

--------------------------------

### Track Agent Sessions with Session ID (TypeScript)

Source: https://docs.thecontext.company/frameworks/ai-sdk/usage

Shows how to group multiple agent runs into a single session by setting a `tcc.sessionId` in the metadata. This is useful for tracking conversations in chatbot interfaces. It requires the 'ai' package.

```typescript
import { generateText } from "ai";

const { text } = await generateText({
  // ...
  experimental_telemetry: {
    isEnabled: true,
    metadata: {
      // use tcc.sessionId to track agent sessions
      "tcc.sessionId": "some-session-id",
    },
  },
});
```

--------------------------------

### Generate and Pass Run ID for User Feedback (TypeScript)

Source: https://docs.thecontext.company/frameworks/mastra/usage

Explains how to generate a unique run ID using `crypto.randomUUID()` and pass it to the agent's `tracingOptions.metadata` under the key `tcc.runId`. This run ID is crucial for associating user feedback with specific agent executions. The generated `runId` should be returned to the client for subsequent feedback submission.

```typescript
import { randomUUID } from "crypto";

// Generate a unique run ID (must be a UUID) before your agent execution
const runId = randomUUID();

const response = await agent.stream(
  // ...
  {
    tracingOptions: {
      metadata: {
        "tcc.runId": runId, // Pass the run ID in metadata
      },
    },
  }
);

// Return the runId to your client
return { response, runId };
```

--------------------------------

### Instrument Next.js Application

Source: https://docs.thecontext.company/frameworks/ai-sdk/local

Registers OpenTelemetry for The Context Company within a Next.js application by calling registerOTelTCC with local mode enabled. This should be placed in an instrumentation.[js|ts] file.

```typescript
export async function register() {
  if (process.env.NEXT_RUNTIME === "nodejs") {
    const { registerOTelTCC } = await import("@contextcompany/otel/nextjs");
    registerOTelTCC({ local: true });
  }
}
```

--------------------------------

### Enable Telemetry for AI SDK Calls

Source: https://docs.thecontext.company/frameworks/vercel

Enables experimental telemetry for AI SDK v5 and later by setting the `experimental_telemetry.isEnabled` flag to `true`. This flag must be included in all AI SDK calls to ensure telemetry data is collected.

```typescript
import { generateText } from "ai";

const result = generateText({
  // ...
  experimental_telemetry: { isEnabled: true }, // required
});
```

--------------------------------

### Add Local Mode Widget to Layout

Source: https://docs.thecontext.company/frameworks/ai-sdk/local

Includes the The Context Company Local Mode widget script in the root layout of a Next.js application. This allows the widget to be available globally.

```tsx
import Script from "next/script";

export default function RootLayout({
  children,
}: {
  children: React.ReactNode
}) {
  return (
    <html lang="en">
      <head>
        {/* add TCC script: */}
        <Script
          crossOrigin="anonymous"
          src="//unpkg.com/@contextcompany/widget/dist/auto.global.js"
        />
      </head>
      <body>{children}</body>
    </html>
  )
}
```

--------------------------------

### Enable Debug Mode for Agent Spans (TypeScript)

Source: https://docs.thecontext.company/frameworks/claude-agent-sdk/usage

Shows how to enable debug mode for logging agent spans by setting `debug: true` within the `tcc` parameter of the `query` function. This helps in inspecting the spans created during agent execution. It utilizes `instrumentClaudeAgent` from '@contextcompany/claude'.

```typescript
import { instrumentClaudeAgent } from "@contextcompany/claude";
import * as claudeSDK from "@anthropic-ai/claude-agent-sdk";

const { query } = instrumentClaudeAgent(claudeSDK);

const result = query({
  // ...
  tcc: {
    debug: true, // Enable debug mode
  },
});
```

--------------------------------

### Instrument Next.js API Calls with registerOTelTCC

Source: https://docs.thecontext.company/frameworks/vercel

Adds instrumentation to Next.js API routes using the `registerOTelTCC` function. This function automatically exports AI SDK-related traces to The Context Company. It requires the `TCC_API_KEY` to be set in the environment.

```typescript
export async function register() {
  if (process.env.NEXT_RUNTIME === "nodejs") {
    const { registerOTelTCC } = await import("@contextcompany/otel/nextjs");
    registerOTelTCC();
  }
}
```

--------------------------------

### Track Agent Sessions with Session ID (TypeScript)

Source: https://docs.thecontext.company/frameworks/mastra/usage

Illustrates how to group multiple agent runs into a single session by setting a `tcc.sessionId` in the `tracingOptions.metadata`. This is commonly used for tracking conversations in chatbot interfaces. The `sessionId` should be a unique string identifier for the session, preferably a UUID.

```typescript
const sessionId = "unique-session-identifier";

await agent.stream(
  // ...
  {
    tracingOptions: {
      metadata: {
        "tcc.sessionId": sessionId, // Track agent sessions
      },
    },
  }
);
```

--------------------------------

### Generate and Pass Run ID for User Feedback (TypeScript)

Source: https://docs.thecontext.company/frameworks/ai-sdk/usage

This snippet shows how to generate a unique run ID using `crypto.randomUUID()` and pass it as custom metadata (`tcc.runId`) during agent execution. This ID is crucial for associating user feedback with specific agent runs. It requires the 'ai' package.

```typescript
import { generateText } from "ai";
import { randomUUID } from "crypto";

// Generate a unique run ID (must be a UUID) before your agent execution
const runId = randomUUID();

const { text } = await generateText({
  // ...
  experimental_telemetry: {
    isEnabled: true,
    metadata: {
      "tcc.runId": runId, // Pass the run ID in metadata
    },
  },
});

// Return the runId to your client
return { text, runId };
```

--------------------------------

### Enable Telemetry for streamText

Source: https://docs.thecontext.company/frameworks/ai-sdk/local

Enables experimental telemetry for the streamText function from the 'ai' library by setting the experimental_telemetry flag to true. This is required for AI SDK v5 and later.

```typescript
import { streamText } from "ai";

const result = streamText({
  // ...
  experimental_telemetry: { isEnabled: true }, // required
});
```

--------------------------------

### Add Custom Metadata to Agent Runs

Source: https://docs.thecontext.company/frameworks/langgraph

Adds custom metadata to LangGraph agent runs by passing it through LangGraph's RunnableConfig. This allows for filtering and analysis of agent runs based on custom properties like user IDs.

```python
result = agent.invoke(
    {"messages": [("user", "Hello!")]},
    {
        "metadata": {
            # e.g. tag this agent run with a user id
            "userId": "4a6b111c-b53a-4d00-a877-67185022ab9e",
        }
    }
)
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.