DemoGPT

// What it does

• Prompt-to-app generation: an instruction and a title become LangChain code wrapped in an interactive Streamlit app
• Staged pipeline — plan, tasks, code snippets, final assembly — with the stages streamed back as JSON from the Python API
• Runs as a local Streamlit interface with the `demogpt` command or as a library via `from demogpt import DemoGPT`
• AgentHub agents: `ToolCallingAgent` and a `ReactAgent` that shows its reasoning and each tool call
• Built-in tools including Tavily search, weather, Wikipedia, Bash, Python, arXiv, YouTube search, Stack Overflow, Wikidata, PubMed and YOLO object detection, plus custom tools via `BaseTool`
• `BaseRAG` retrieval over PDF, TXT, CSV and JSON files with Chroma, Pinecone or FAISS vector stores, usable on its own or as an agent tool

// Getting started

DemoGPT installs from PyPI. The generator uses an OpenAI model by default, so set `OPENAI_API_KEY` in your environment (or pass `openai_api_key` to the Python class).

pip install demogpt

# or from source
git clone https://github.com/melih-unsal/DemoGPT.git
cd DemoGPT
pip install .

demogpt

# from a source checkout
streamlit run demogpt/app.py

from demogpt import DemoGPT

agent = DemoGPT(model_name="gpt-3.5-turbo")
instruction = "Your instruction here"
title = "Your title here"
code = ""
for phase in agent(instruction=instruction, title=title):
    print(phase)
    if phase["done"]:
        code = phase["code"]
print(code)

from demogpt_agenthub.agents import ToolCallingAgent
from demogpt_agenthub.llms import OpenAIChatModel
from demogpt_agenthub.tools import TavilySearchTool, WeatherTool

llm = OpenAIChatModel(model_name="gpt-4o-mini")
agent = ToolCallingAgent(tools=[TavilySearchTool(), WeatherTool()], llm=llm, verbose=True)
print(agent.run("What's the weather like in New York today?"))

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Turn an idea for a small LLM tool into a clickable Streamlit prototype without writing the LangChain code by hand
• Generate a starting LangChain program from Python and edit the returned code instead of beginning from a blank file
• Build a tool-using agent that searches the web, runs Python or Bash, and checks the weather, with its reasoning printed step by step
• Answer questions over your own PDFs, CSVs or JSON files with a RAG pipeline that an agent can call alongside other tools

// How DemoGPT compares

DemoGPT alongside other open-source low-code & no-code builders tools AI/TLDR tracks, ranked by GitHub stars.