EuniAI/Prometheus

🧠 Prometheus: A Knowledge-Graph-Driven 🤖 AI Agent that maps 🗺, understands 🧩, and repairs 🛠 complex codebases — not by guessing, but by reasoning. ⚡

What it solves

Prometheus is designed to move beyond simple prototype generation to create verifiable, production-ready software. It addresses the unreliability of "vibe coding" by automating the complex, long-horizon tasks of software engineering, such as reproducing bugs, implementing features, and navigating large, multilingual codebases.

How it works

The platform uses a hierarchical multi-agent system built on LangGraph state machines to orchestrate specialized AI agents. It employs a Neo4j-powered unified knowledge graph (using Tree-sitter for AST and semantic representation) to provide deep repository-wide understanding. The workflow is divided into specialized pipelines:

  • Bug Pipeline: Classifies issues, reproduces bugs in Docker-isolated environments, and generates validated patches.
  • Feature Pipeline: Analyzes feature requests, plans implementation, and generates code with optional regression testing.
  • Question Pipeline: Performs graph-based semantic search and tool-augmented analysis to answer natural language queries about the codebase.

Who it’s for

It is intended for software engineers and organizations looking to automate the maintenance and evolution of complex repositories through autonomous AI agents.

Highlights

  • Full-Cycle Automation: Implements a complete detect-reproduce-repair-verify (DRRV) loop for bug fixing.
  • Knowledge Graph Integration: Uses a structured semantic representation of code to enable better cross-file reasoning than standard LLM interfaces.
  • High Performance: Achieved top rankings on the SWE-bench leaderboard for automated software engineering.
  • Isolated Execution: Uses Docker containers to ensure that testing and validation of generated patches are safe and isolated.

Related

  • Project
    zi-yue-1129/DATAGEN

    An AI-powered data analysis and research platform that uses a multi-agent system to automate hypothesis generation, data processing, visualization, and report writing.

  • Project
    larlarua/AutoCVE

    An automated vulnerability discovery system that uses a multi-agent AI framework to screen projects, audit source code, and generate CVE reports.

  • Project
    agentenv/agentflow

    An orchestration framework for coordinating multi-agent LLM workflows using dependency graphs, featuring parallel fanout, iterative loops, and isolated execution on Docker, VMs, or cloud targets.

  • Project
    colbymchenry/codegraph

    CodeGraph is a Rust‑based local tool that builds a full‑code knowledge graph for any project (supports dozens of languages) and integrates it with AI coding assistants (Claude Code, Cursor, Copilot, Gemini, etc.). The graph lets agents retrieve exact symbols and call‑paths in a single request, cutting tool calls, file reads, token usage and cost dramatically. Install via a one‑line script, run `codegraph init` per repo, and the CLI auto‑updates the graph as you edit. A browser UI (`codegraph ui`) visualises the graph, and the CLI can uninstall itself cleanly.

  • Project
    promptise-com/Foundry

    A full-stack agentic engineering framework that provides a unified system for building, reasoning, and deploying secure, multi-tenant AI agents with integrated tools and runtime.