deepground 0.8.2

Deepground - A Python library that empowers LLM agents with real-time data access from both the clear and dark net.

Deepground

Deepground – Clearnet + Darknet Search & Fetcher

Deepground - this is a Python library that empowers LLM agents with real-time data access from buth the clearnet and darknet.
This library is design for developers, hackers, pentesters, researchers, and AI agents who need real-time datas from both the clearnet and darknet to their CLI.

Overview

The deepground project was developed by evilfeonix and designed to evaluate the performance of llm and their applications. It gives your LLM the power to access and monitor real-time data from the net for free without limitation or restriction.

Features

• Support sync / async(parallel fetching) mode, depending on your workflow.
• Fetch context and extract sources (raw HTML/code/text) from URLs.
• Integrate seamlessly with LangChain and other LLM frameworks.
• Searches across clearnet/darknet sources with ease.
• Built-in caching + logging
• Error handling

Installation

pip install deepground

For dev:

git clone https://github.com/evilfeonix/deepground.git
cd deepground
pip install -e .
python3 run.py

Usage

Simple Usage:

from deepground.core import Grounder

g = Grounder()
print(g.search("latest AI news"))

g_tor = Grounder(use_tor=True)
print(g_tor.dark_search("leaked databases"))

Synchronous:

from deepground.core import Grounder


# Clearnet search
g = Grounder(use_tor=False)
r = g.search("Python hacking", limit=3)
print(r)

# Darknet search
g_tor = Grounder(use_tor=True)
d_r = g_tor.dark_search("market", limit=3)
print(d_r)

# Fetch readable content (page)
p = g.fetch_context("https://example.com")
print(p["content"])

# Fetch row source (code/text)
s = g_tor.fetch_source("https://somesite.onion/ABCDEF.txt")
print(s["source"])

Asynchronous:

import asyncio
from deepground.core import GrounderAsync

async def main():
    async with GrounderAsync(use_tor=False) as g:
        results = await g.search("cybersecurity news", limit=3)
        print(results)

        source = await g.multi-fetch("https://example.com,https://example.net", context=False)
        print(source["source"][:300])

asyncio.run(main())

LangChain Integration:

from deepground.core import GrounderTool

tool = GrounderTool(use_tor=True)

print(tool._run("search:latest AI news"))
print(tool._run("dark_search:proxies combo list"))
print(tool._run("fetch_context:https://github.com"))
print(tool._run("fetch_source:https://somesite.onion"))
print(tool._run("multi-fetch:https://ss1.onion,https://ss2.onion"))

Langchain Agent

Feed results directly into your LLM agents for grounded, real-world context:

from deepground.core import Grounder

# Clearnet search
g = Grounder()
res = g.search("naija latest tech news")
ctx = g.fetch_context("https://example.com/article")

# Darknet search with Tor
g_tor = Grounder(use_tor=True)
d_res = g_tor.dark_search("fresh proxies list")
d_ctx = g_tor.fetch_context("http://somesite.onion/article")

# Feed into LLM

from langchain_openai import ChatOpenAI
llm = ChatOpenAI(model="gpt-4o-mini", temperature=0)

answer = llm.invoke(f"Summarize:\n\n{ctx}")
print(answer)

For darknet features:

TOR running locally (socks5://127.0.0.1:9050)

pip install requests[socks]

Logging & Caching

Cache stored at: ~/.deepground/cache
Log file: ~/.deepground/deepground.log

Documentation

Detailed documentation includeing full guides, and examples can be found at our official website https://evilfeonix.eu.org

Contrubuting

Contributors are welcome to this journey, feel free to :

• Report bugs via github issues.
• Suggest feature with use cases.
• Submit PRs with clear description.

Disclaimer

This library is for educational and research purposes only. Authors will never ever be responsible for any misuse or damage cause by this project.

LICENSE

Use MIT license:

MIT License
Copyright (c) 2025 Evilfeonix
Permission is hereby granted, free of charge...

Powered by

This project is made posibly thanks to those infrastructure with their generous supports.