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

# Import Agents

> This guide will walk you through importing specific AI Agents from the hub.

## Overview

You can access mkinf hub Agents throught the [mkinf python SDK](https://pypi.org/project/mkinf/) or via REST API calls.

### 1. Find an AI Agent

Browse available AI Agents at [mkinf hub](https://hub.mkinf.io/) and select an agent that matches your use case

{" "}

<img src="https://mintcdn.com/thecomputecompany/Kei6u5YuNTROU2k7/images/repository-card.png?fit=max&auto=format&n=Kei6u5YuNTROU2k7&q=85&s=a112c85ef4734c7012a736908d521c2c" alt="Repository card showing an AI Agent" width="2602" height="256" data-path="images/repository-card.png" />

### 2. Copy the pull command

Check the "Use Agent" section of your chosen repository for import instructions

{" "}

<img src="https://mintcdn.com/thecomputecompany/Kei6u5YuNTROU2k7/images/use-agent-card.png?fit=max&auto=format&n=Kei6u5YuNTROU2k7&q=85&s=2b4d6064e4efd9158a4305becd64e90a" alt="Card showing how to import the AI Agent" width="2426" height="1740" data-path="images/use-agent-card.png" />

## Python

### Import the AI Agent

To import the agents in our platform, follow this structure:

```python theme={null}
from mkinf import hub as mh
tools = mh.pull(["owner/repository"])
```

Some tools require environment variables, including API keys, which must be created and set before use.

```python theme={null}
from mkinf import hub as mh
tools = mh.pull(
    ["ScrapeGraphAI/scrapegraphai"],
    env={
        "SCRAPEGRAPH_LLM_MODEL": "openai/gpt-4o-mini",
        "SCRAPEGRAPH_LLM_API_KEY": os.getenv("OPENAI_API_KEY")
    },
    timeout=120
)
```

You can even set an execution timeout, the default execution timeout is **60 seconds**.

<Note>Remember to configure any required environment variables specified in the agent's documentation.</Note>

<Warning>Currently, mkinf tools are compatible with LangChain chains and graphs. Support for other frameworks like CrewAI, AutoGen, and SmolAgents is coming soon.</Warning>

## REST API

### Request execution

Agents actions can also be executed directly via API call.

```bash theme={null}
curl -X POST "https://run.mkinf.io/v0.1/<REPO_OWNER>/<REPO_NAME>:<REPO_VERSION>/<REPO_ACTION>" \
  -H "Authorization: Bearer $MKINF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "args": { "your_arg_key": "your_arg_value" },
    "env": { "your_env_key": "your_env_value" },
    "timeout": 60
  }'
```

Note that you have also to include the input arguments, you can them in the **input schema** of each action.

{" "}

<img src="https://mintcdn.com/thecomputecompany/Kei6u5YuNTROU2k7/images/action.png?fit=max&auto=format&n=Kei6u5YuNTROU2k7&q=85&s=34767d068cfc29bfd147080fb5aa4e05" alt="Card showing AI Agent input and output schema" width="1646" height="1190" data-path="images/action.png" />

## Next Steps

* Check out our [SDK repository](https://github.com/mkinf-io/mkinf-py/tree/main/example) for implementation samples
* Join our [Slack community](https://join.slack.com/t/mkinf-community/shared_invite/zt-2wa3mqqsx-27qHzCok~BCdZTowifer~g) for support