LLAMOSC stands for LLM Agent-Based Modeling for Open Source Communities. The name encapsulates the core components and objectives of the framework:
LLM: Large Language Models, the backbone of our intelligent agents, providing them with advanced perception, reasoning, and decision-making capabilities.
ABM: Agent-Based Modeling, the methodological approach we use to simulate the interactions and dynamics within open-source communities.
OSC: Open Source Communities, the focus of our simulations, aiming to enhance sustainability, collaboration, and engagement within these communities.
Key Components and Features implemented so far
Environment:
GitHub (CodeSpace Environment): A simulated code space (powered by AutoCodeRover) where agents contribute to open-source projects, mirroring real-world GitHub repositories.
Environment Variables:
Issues: Issues are categorized with varying difficulty from 1 to 5, resembling real-life variation from "Good First Issues" to "Expert Required Issues", assigned based on the contributor agent's experience level.
Agents:
Coding Ability: Skill level in writing and reviewing code, (powered by AutoCodeRover).
Agent Variables (Internal States):
Experience Level: Categorized from 1 to 5 resembling real-life contributors ranging from Novice to Experienced.
Framework Capabilities
LLAMOSC integrates multiple capabilities to automate and optimize open-source community activities:
AutoCodeRover Integration:
Combines LLMs with advanced code navigation and debugging capabilities.
Operates in two stages: Context retrieval and Patch generation.
Features Program Structure Aware code search APIs to navigate codebases and generate patches.
Automated Pull Request Lifecycle:
ContributorAgent: Automates issue identification, solution proposal, and pull request creation.
MaintainerAgent: Automates pull request review and merging based on predefined criteria.
Utilizes Docker for environment consistency and isolation.
Implements diff extraction and automated pull request submission (powered by AutoCodeRover).
LLAMOSC stands for LLM Agent-Based Modeling for Open Source Communities. The name encapsulates the core components and objectives of the framework:
Key Components and Features implemented so far
Environment:
Environment Variables:
Agents:
Agent Variables (Internal States):
Framework Capabilities
LLAMOSC integrates multiple capabilities to automate and optimize open-source community activities:
AutoCodeRover Integration:
Automated Pull Request Lifecycle:
Multi-Agent Decision Making Algorithms:
Accomplishments of LLAMOSC so far
Preliminary Design and Implementation:
Automated Pull Request Lifecycle:
Multi-Agent Decision Making: