Skip to content

LangGraph based Multi-agent Collaboration workflow based on Plan-and-execute with tool-use access and skeleton-of-code.

Notifications You must be signed in to change notification settings

rvencharla/Devin2.0_llm_agent

Repository files navigation

Team 5 - Ashby Prize in Computational Science Hackathon

Introducing the pinnacle of artificial intelligence-driven engineering marvel: our bespoke AI Software Engineer. This cutting-edge system operates at the forefront of technological innovation, seamlessly interfacing with users to decipher high-level requirements into a meticulously orchestrated symphony of planning and task delegation.

Harnessing the power of advanced algorithms and machine learning, our AI Software Engineer embarks on a journey of unparalleled complexity, navigating through the intricate landscape of software development with unparalleled finesse. Upon receiving the user's directive, it orchestrates a grand ballet of planning, meticulously dissecting the overarching goal into a cascading cascade of tasks and subtasks, each meticulously tailored to optimize efficiency and effectiveness.

These subtasks, akin to the finest gears within a Swiss timepiece, engage in the delicate art of coding, weaving intricate webs of logic and functionality with unparalleled precision. Through an intricate dance of abstraction and implementation, our AI Software Engineer breathes life into the digital realm, transforming abstract concepts into tangible realities with unparalleled elegance and sophistication.

As the culmination of its efforts, our AI Software Engineer presents the final output—a masterpiece of computational ingenuity, meticulously crafted to meet and exceed the user's expectations. With its unwavering dedication to excellence and its insatiable thirst for innovation, our AI Software Engineer stands as a beacon of progress in the ever-evolving landscape of software engineering.

Environment Variables

To run this project, you will need to add the following environment variables to your .env file

  • AZURE_OPENAI_API_KEY
  • AZURE_OPENAI_ENDPOINT
  • AZURE_OPENAI_MODEL
  • AZURE_OPENAI_API_VERSION
  • WOLFRAM_LLM_API_KEY
  • SEARCHAPI_API_KEY
  • E2B_API_KEY
  • BEARLY_API_KEY

Run Locally

Clone the project

  git clone https://github.com/rvencharla/Devin2.0.git

Go to the project directory

  cd team5

Install dependencies

  pip install -r requirements.txt

Create a .env file and populate it with the API Keys

  cat .env

Run the App

  streamlit run main.py

Architecture

App Screenshot Screenshot 2024-04-23 at 9 46 21 PM

  • The user prompts the application with the project requirements and outcomes.
  • The bot divides the tasks into a series of sub-problems.
  • Each of the sub-problem is assigned to separate agents fo solving.
  • A test agent generates test cases for each of the functions generated by the coder agents.
  • If the tests are sucessfull, then the code is pushed to another agents which stitches up multiple pieces of code else, the code is corrected for those generated testcases.

Screenshots

App Screenshot

App Screenshot

App Screenshot

App Screenshot

App Screenshot

About

LangGraph based Multi-agent Collaboration workflow based on Plan-and-execute with tool-use access and skeleton-of-code.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages