> ## Documentation Index
> Fetch the complete documentation index at: https://docs.twexapi.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Microsoft Agent Framework

> Use Twexapi MCP from Microsoft agent workflows.

Use Microsoft Agent Framework when you are building .NET or Python agent workflows and want Twexapi exposed as a remote tool source.

## Prerequisites

* A Twexapi API key
* Python 3.10+ or a .NET agent host with MCP support
* A model configured for your Microsoft agent runtime
* Streamable HTTP MCP support in the agent host or a small bridge process

## Install

```bash theme={null}
pip install agent-framework mcp python-dotenv
```

## Full example

```python theme={null}
import asyncio
import os
from pathlib import Path

from agent_framework import ChatAgent, MCPStreamableHTTPTool
from agent_framework.openai import OpenAIChatClient
from dotenv import load_dotenv


async def main():
    load_dotenv()

    mcp_tool = MCPStreamableHTTPTool(
        name="twexapi",
        url="https://api.twexapi.io/mcp",
        headers={"x-api-key": os.environ["TWEXAPI_API_KEY"]},
        description="Twexapi X/Twitter API tools exposed through MCP",
    )

    async with mcp_tool:
        agent = ChatAgent(
            chat_client=OpenAIChatClient(model_id="gpt-4o"),
            name="twexapi_agent",
            instructions=(
                "You help users interact with X/Twitter via Twexapi. "
                "Call explore first, then twexapi_request. Preserve IDs and cursors."
            ),
            tools=[mcp_tool],
        )

        response = await agent.run(
            "Search for tweets about AI agents. Return compact JSON with "
            "route_used, tweets[{tweet_id,text,author_username,created_at}], "
            "has_more, and next_cursor."
        )

        Path("twexapi-agent-framework-handoff.json").write_text(
            response.text,
            encoding="utf-8",
        )


asyncio.run(main())
```

## MCP connection JSON

```json theme={null}
{
  "name": "twexapi",
  "transport": "streamable-http",
  "url": "https://api.twexapi.io/mcp",
  "headers": {
    "x-api-key": "twexapi_YOUR_KEY_HERE"
  }
}
```

## System instruction

```txt theme={null}
You have access to Twexapi MCP tools.
Call explore before twexapi_request.
Use only relative paths returned by explore.
Return tweet_id, user_id, author_username, route_used, has_more, and next_cursor.
Ask for confirmation before read_only: false actions.
```

## Example task

```txt theme={null}
Research current X/Twitter discussion about "AI coding agents".
Use Twexapi MCP, collect 25 recent tweets, group them by theme, and return compact JSON.
Include tweet_id, author_username, text, created_at, engagement metrics, route_used, has_more, and next_cursor.
```

## Handoff checklist

* Store `tweet_id`, `text`, `author_username`, `created_at`, and engagement metrics.
* Store `user_id`, `username`, and verification fields for profile workflows.
* Store `has_more`, `next_cursor`, `method`, `path`, and `route_used`.
* Stop before write actions unless a human approves the exact action.

## Environment variables

```txt .env theme={null}
TWEXAPI_API_KEY=twexapi_YOUR_KEY_HERE
OPENAI_API_KEY=sk-...
```

## Production guidance

* Use a per-environment API key instead of embedding keys in agent prompts.
* Log MCP tool calls and the returned `path`.
* Treat any write-capable endpoint as a human-approved step.
* Persist cursors outside the conversation state.

## Package versions

| Package           | Version |
| ----------------- | ------- |
| `agent-framework` | `0.2+`  |
| `mcp`             | `1.9+`  |