Metadata-Version: 2.4
Name: agentai
Version: 0.1.3
Summary: A Python client for the Agent.ai API
Home-page: https://github.com/OnStartups/python_sdk
Author: Andrei Oprisan
Author-email: andrei@agent.ai
Project-URL: Documentation, https://github.com/OnStartups/python_sdk#readme
Project-URL: Bug Tracker, https://github.com/OnStartups/python_sdk/issues
Keywords: agent.ai,ai,llm,api,client,chatbot,automation
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: requests>=2.25.0
Dynamic: author
Dynamic: author-email
Dynamic: classifier
Dynamic: description
Dynamic: description-content-type
Dynamic: home-page
Dynamic: keywords
Dynamic: license-file
Dynamic: project-url
Dynamic: requires-dist
Dynamic: requires-python
Dynamic: summary

# agent.ai Python SDK

[![PyPI version](https://badge.fury.io/py/agentai.svg)](https://badge.fury.io/py/agentai)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)

A simple and powerful Python library for interacting with the [Agent.ai Actions API](https://agent.ai/actions).

## Features

- 🔑 **Easy Authentication** - Use environment variables or pass token directly
- 🚀 **Convenience Methods** - Simple methods for common actions
- 🛡️ **Robust Error Handling** - Clear error messages from the API
- 📝 **Well Documented** - Comprehensive examples and docstrings
- 🔧 **Flexible** - Use convenience methods or raw action calls
- 🤖 **LLM-Friendly** - Includes [llm.txt](llm.txt) specification for AI assistants

## Installation

```bash
pip install agentai
```

## Quick Start

### 1. Get Your API Key

Sign up at [Agent.ai](https://agent.ai) and get your Bearer token from [account settings](https://agent.ai/user/settings#credits).

### 2. Set Up Authentication

**Option A: Environment Variable (Recommended)**
```bash
export AGENTAI_API_KEY="your_api_key_here"
```

**Option B: Pass Token Directly**
```python
from agentai import AgentAiClient
client = AgentAiClient(bearer_token="your_api_key_here")
```

### 3. Start Using the API

```python
from agentai import AgentAiClient

client = AgentAiClient()  # Uses AGENTAI_API_KEY env var

# Chat with an LLM
response = client.chat("What is machine learning?")
if response['status'] == 200:
    print(response['results'])
```

## Usage Examples

### Chat with LLMs

```python
from agentai import AgentAiClient

client = AgentAiClient()

# Simple chat
response = client.chat("Explain quantum computing in simple terms")
print(response['results'])

# Specify a model
response = client.chat(
    prompt="Write a haiku about coding",
    model="claude-sonnet"  # Options: gpt4o, gpt4o-mini, claude-sonnet, claude-haiku, 
                           # gemini-pro, gemini-flash, llama-70b, llama-8b, deepseek
)
print(response['results'])
```

### Web Scraping

```python
# Get text from a webpage
response = client.grab_web_text("https://example.com")
if response['status'] == 200:
    print(response['results'])

# Take a screenshot
response = client.grab_web_screenshot("https://example.com", full_page=True)
if response['status'] == 200:
    print(f"Screenshot URL: {response['results']}")
```

### YouTube

```python
# Get video transcript
response = client.get_youtube_transcript("https://www.youtube.com/watch?v=dQw4w9WgXcQ")
print(response['results'])

# Search YouTube
response = client.search_youtube("python tutorials", max_results=5)
for video in response['results']:
    print(video['title'])

# Get channel info
response = client.get_youtube_channel("https://www.youtube.com/@PythonTutorials")
print(response['results'])
```

### Social Media

```python
# LinkedIn
response = client.get_linkedin_profile("company/hubspot")
print(response['results'])

# Twitter/X
response = client.get_recent_tweets("elonmusk", count=5)
for tweet in response['results']:
    print(tweet['text'])

# Bluesky
response = client.get_bluesky_posts("user.bsky.social", count=10)
print(response['results'])

# Instagram
response = client.get_instagram_profile("instagram")
print(response['results'])
```

### News & Search

```python
# Google News
response = client.get_google_news(
    query="artificial intelligence",
    date_range="7d",  # Options: 1d, 7d, 30d, 1y
    location="US"
)
for article in response['results'].get('organic_results', []):
    print(f"- {article['title']}")

# Web Search
response = client.search_web("best python libraries 2024", num_results=10)
print(response['results'])
```

### Company Intelligence

```python
# Get company info from domain
response = client.get_company_info("hubspot.com")
if response['status'] == 200:
    company = response['results']
    print(f"Name: {company.get('name')}")
    print(f"Industry: {company.get('category', {}).get('industry')}")
    print(f"Employees: {company.get('metrics', {}).get('employeesRange')}")
```

### Image Generation

```python
response = client.generate_image(
    prompt="A futuristic city with flying cars",
    model="DALL-E 3",        # or "Stable Diffusion"
    style="digital art",
    aspect_ratio="16:9"      # Options: 1:1, 16:9, 9:16, 4:3, 3:4
)
if response['status'] == 200:
    print(f"Image URL: {response['results']['images'][0]['url']}")
```

### Text to Speech

```python
response = client.text_to_audio(
    text="Hello, welcome to agent.ai!",
    voice="alloy"
)
if response['status'] == 200:
    print(f"Audio URL: {response['results']}")
```

### File Conversion

```python
# Check available conversion formats
response = client.get_convert_options("pdf")
print(f"PDF can be converted to: {response['results']}")

# Convert a file
response = client.convert_file(
    file_url="https://example.com/document.pdf",
    target_format="txt"
)
if response['status'] == 200:
    print(f"Converted file: {response['results']}")
```

### Database Variables

```python
# Store a variable
client.store_variable("user_preference", "dark_mode")

# Retrieve a variable
response = client.get_variable("user_preference")
print(response['results'])
```

### REST API Calls

```python
response = client.rest_call(
    url="https://api.example.com/data",
    method="POST",
    headers={"X-Custom-Header": "value"},
    body={"key": "value"}
)
print(response['results'])
```

### Using Raw Actions

For full flexibility, use the `action()` method directly:

```python
response = client.action(
    action_id="getGoogleNews",
    params={
        "query": "AI news",
        "date_range": "7d",
        "location": "Boston"
    }
)
print(response['results'])
```

## Available Actions

| Action ID | Description | Convenience Method |
|-----------|-------------|-------------------|
| `grabWebText` | Get text from a webpage | `grab_web_text()` |
| `grabWebScreenshot` | Screenshot a webpage | `grab_web_screenshot()` |
| `getYoutubeTranscript` | Get YouTube video transcript | `get_youtube_transcript()` |
| `getYoutubeChannel` | Get YouTube channel info | `get_youtube_channel()` |
| `runYoutubeSearch` | Search YouTube | `search_youtube()` |
| `getGoogleNews` | Get Google News articles | `get_google_news()` |
| `getSearchResults` | Web search results | `search_web()` |
| `getLinkedinProfile` | Get LinkedIn profile | `get_linkedin_profile()` |
| `getLinkedinActivity` | Get LinkedIn posts | `get_linkedin_activity()` |
| `getCompanyObject` | Company intelligence | `get_company_info()` |
| `getTwitterUsers` | Twitter user info | `get_twitter_user()` |
| `getRecentTweets` | Get recent tweets | `get_recent_tweets()` |
| `getBlueskyPosts` | Get Bluesky posts | `get_bluesky_posts()` |
| `searchBlueskyPosts` | Search Bluesky | `search_bluesky()` |
| `getInstagramProfile` | Instagram profile | `get_instagram_profile()` |
| `invokeLlm` | Chat with LLM | `chat()` |
| `generateImage` | Generate images | `generate_image()` |
| `outputAudio` | Text to speech | `text_to_audio()` |
| `convertFile` | Convert files | `convert_file()` |
| `convertFileOptions` | Get conversion options | `get_convert_options()` |
| `storeVariableToDatabase` | Store variable | `store_variable()` |
| `getVariableFromDatabase` | Get variable | `get_variable()` |
| `invokeAgent` | Call another agent | `invoke_agent()` |
| `restCall` | Make REST API call | `rest_call()` |

## Error Handling

All methods return a dictionary with the following structure:

```python
{
    "status": 200,           # HTTP status code (or None for connection errors)
    "error": None,           # Error message (or None if successful)
    "results": {...},        # API response data (or None if error)
    "metadata": {...}        # Additional metadata (if available)
}
```

### Example Error Handling

```python
response = client.chat("Hello!")

if response['status'] == 200:
    print(f"Success: {response['results']}")
elif response['status'] == 403:
    print("Authentication failed. Check your API key.")
elif response['status'] == 429:
    print("Rate limited. Please wait and try again.")
elif response['status'] is None:
    print(f"Connection error: {response['error']}")
else:
    print(f"Error {response['status']}: {response['error']}")
```

## Examples

See the `examples/` directory for complete, runnable examples:

- **`interactive_chat.py`** - Chat with different LLM models
- **`youtube_summary.py`** - Summarize YouTube videos
- **`lead_enrichment.py`** - Enrich leads with company & LinkedIn data
- **`brand_news_monitor.py`** - Monitor Google News for brand mentions
- **`competitor_analysis.py`** - Analyze competitor websites & social presence
- **`social_media_content.py`** - Generate social posts with images
- **`faq_chatbot.py`** - Build a chatbot from website content
- **`sales_email_subject.py`** - Generate email subject lines
- **`file_conversion.py`** - Convert files between formats
- **`website_monitor.py`** - Monitor websites for changes
- **`web_search.py`** - Search the web and analyze results
- **`bluesky_monitor.py`** - Monitor Bluesky posts

## LLM Models

Available models for the `chat()` method:

| Model | Description |
|-------|-------------|
| `gpt4o` | OpenAI GPT-4o (default) |
| `gpt4o-mini` | OpenAI GPT-4o Mini (faster, cheaper) |
| `claude-sonnet` | Anthropic Claude Sonnet |
| `claude-haiku` | Anthropic Claude Haiku (faster) |
| `gemini-pro` | Google Gemini Pro |
| `gemini-flash` | Google Gemini Flash (faster) |
| `llama-70b` | Meta Llama 70B |
| `llama-8b` | Meta Llama 8B (faster) |
| `deepseek` | DeepSeek |

## For AI Assistants (llm.txt)

This package includes an [`llm.txt`](llm.txt) file - a comprehensive specification designed for LLM assistants (like ChatGPT, Claude, Copilot, etc.) to understand and use this SDK effectively.

The `llm.txt` file contains:
- Complete API reference with all methods and parameters
- Response format documentation
- Common use case examples (research assistant, social monitor, content generator, etc.)
- Error handling patterns
- Tips for optimal usage

**For AI coding assistants:** Reference the `llm.txt` file when helping users write code with this SDK.

## Contributing

Contributions are welcome! Please feel free to submit pull requests or open issues for feature requests or bug reports.

## License

This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.

## Links

- [Agent.ai Website](https://agent.ai)
- [API Documentation](https://agent.ai/actions)
- [Get API Key](https://agent.ai/user/settings#credits)
- [GitHub Repository](https://github.com/OnStartups/python_sdk)
- [LLM Specification](llm.txt) - For AI assistants
