Skip to content

Latest commit

 

History

History
73 lines (53 loc) · 2.49 KB

tools.md

File metadata and controls

73 lines (53 loc) · 2.49 KB
Raw

Gmail

Before using the Gmail tool, you need to set up SMTP access. Please follow the instructions in this blog post or this Chinese blog to obtain a device password.

Use Case

from miniagent.tools import EmailTool

tool = EmailTool(user_email='your_email', password='your_device_password', recipient=["your_recipient"])

tool.invoke(subject="Hello World!", contents="Hello World!", email="Other recipients desired to send")

Arxiv

This tool allows you to search for relevant papers on Arxiv.

Use Case

from miniagent.tools import ArxivTool

arxivtool = ArxivTool(sort_criterion="submittedDate")
output = arxivtool.invoke("language agent")

Search Engine

NOTE: We use the Search-Engines-Scraper to get the searched results. However, the searched results are very limited and not the intact information. A searched sample is:

{'host': 'healthline.com',                                                                                                                                             
 'link': 'https://www.healthline.com/health/staying-healthy',
 'text': 'Learn how to improve your physical and mental well-being with '
         'science-backed advice on exercise, nutrition, smoking, sleep, '
         'hydration, and more. Find out how to lower your risk of chronic '
         'diseases, improve your mood, and live a long, productive life.',
 'title': 'Staying Healthy: Top 10 Tips for Good Health'}

The text is a short description for the url. So, it is better to scraped detailed information from the link or url according to the interested content of the LLM.

PDFReader

This tool is designed to parse PDF files into text format.

Use Case

from miniagent.tools import PDFReaderTool

pdftool = PDFReaderTool()
filepath = "examples/example.pdf"
parsed_text = pdftool.invoke(filepath)

ScrapTool

The ScrapTool is intended for extracting and returning textual content from a specified URL.

Use Case

from miniagent.tools import ScrapTool

scraptool = ScrapTool()
output = scraptool.invoke(url="https://www.langchain.com")
print(output)

This tool will return the main content of a given HTML page. It can be extended to create a ScrapAgent that scrapes raw text from a URL and summarizes it.