> ## 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.

# CrewAI

> Build multi-agent X/Twitter research crews with Twexapi MCP.

Use CrewAI when you want separate agents for collection, analysis, and reporting while sharing one Twexapi MCP connection.

## Prerequisites

* Python 3.10+
* A Twexapi API key from the [dashboard](https://twexapi.io/dashboard)
* An LLM API key supported by CrewAI

## Install

```bash theme={null}
pip install crewai python-dotenv
```

## Full example

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

from crewai import Agent, Crew, Process, Task
from crewai.mcp import MCPServerHTTP
from dotenv import load_dotenv

load_dotenv()

twexapi_mcp = MCPServerHTTP(
    url="https://api.twexapi.io/mcp",
    headers={"x-api-key": os.environ["TWEXAPI_API_KEY"]},
)

researcher = Agent(
    role="X Research Analyst",
    goal="Collect structured public X/Twitter data with Twexapi MCP",
    backstory="A careful analyst who preserves IDs, cursors, and route names.",
    mcps=[twexapi_mcp],
)

task = Task(
    description=(
        "Search recent tweets about AI agents. Use explore first. Return compact JSON "
        "with route_used, tweets[{tweet_id,text,author_username,created_at}], "
        "has_more, next_cursor, and key influencers."
    ),
    expected_output="Compact JSON with tweet rows, pagination state, and influencer notes.",
    agent=researcher,
)

crew = Crew(
    agents=[researcher],
    tasks=[task],
    process=Process.sequential,
    verbose=True,
)

result = crew.kickoff()
Path("twexapi-crewai-handoff.json").write_text(str(result), encoding="utf-8")
```

## Multi-agent crew

Give only the data-collection agents Twexapi access. Downstream agents can work from the handoff JSON.

```python theme={null}
analyst = Agent(
    role="Engagement Analyst",
    goal="Find patterns in tweet rows and profile rows",
    backstory="Transforms structured social data into concise findings.",
)

writer = Agent(
    role="Report Writer",
    goal="Write a short executive summary from the analyst output",
    backstory="Creates readable summaries without inventing missing metrics.",
)
```

## Handoff checklist

* Preserve `tweet_id`, `user_id`, `author_username`, `created_at`, and source query.
* Preserve `has_more` and `next_cursor` for every paginated call.
* Store `route_used`, `method`, and `path` so the workflow can be audited.
* Require human approval before any `read_only: false` action.

## Tool filtering

When a CrewAI role should only discover endpoints, limit that agent to `explore`. Give `twexapi_request` only to agents that are allowed to make authenticated API calls.

```python theme={null}
from crewai.mcp import MCPServerHTTP
from crewai.mcp.filters import create_static_tool_filter

read_only_catalog = MCPServerHTTP(
    url="https://api.twexapi.io/mcp",
    headers={"x-api-key": os.environ["TWEXAPI_API_KEY"]},
    tool_filter=create_static_tool_filter(
        allowed_tool_names=["explore"],
    ),
)
```

For production research crews, keep write-capable work in a separate task that asks for explicit approval before any `read_only: false` call.

## Environment variables

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

```python theme={null}
import os
from crewai.mcp import MCPServerHTTP

twexapi_mcp = MCPServerHTTP(
    url="https://api.twexapi.io/mcp",
    headers={"x-api-key": os.environ["TWEXAPI_API_KEY"]},
)
```

## Package versions

| Package  | Version |
| -------- | ------- |
| `crewai` | `1.0+`  |
| `mcp`    | `1.9+`  |