aegra  by ibbybuilds

AI agent backend for LangGraph Platform users seeking self-hosted control

Created 3 months ago
296 stars

Top 89.4% on SourcePulse

GitHubView on GitHub
Project Summary

Summary

Aegra provides a self-hosted, open-source alternative to platforms like LangGraph, enabling developers to build and deploy AI agent backends with full control over their infrastructure. Built with FastAPI and PostgreSQL, it offers zero vendor lock-in, data sovereignty, and cost optimization for teams requiring custom deployments and complete ownership of their agent orchestration.

How It Works

Aegra functions as an Agent Protocol-compliant HTTP API layer powered by FastAPI, interfacing with the LangGraph SDK for state management and graph execution. Agent run data and persistent conversations are stored in a user-managed PostgreSQL database. This architecture allows seamless integration with existing LangGraph Client SDK applications while ensuring data control and avoiding vendor lock-in through its open-source nature and Apache 2.0 license.

Quick Start & Requirements

  • Prerequisites: Python 3.11+, Docker (for PostgreSQL), uv package manager.
  • Primary Command: docker compose up aegra after cloning the repository and setting up dependencies with uv sync.
  • Verification: Health check at http://localhost:8000/health and interactive API docs at http://localhost:8000/docs.
  • Resources: Official quick-start guide within the README.

Highlighted Details

  • Agent Protocol Compliant: Implements the open-source standard for serving LLM agents in production.
  • LangGraph SDK Compatibility: Existing LangGraph applications connect and run without modification.
  • Self-Hosted & Zero Lock-in: Full control over infrastructure, data, and authentication (JWT/OAuth/Firebase/NoAuth).
  • Human-in-the-Loop Support: Enables interactive agent workflows with approval gates.
  • Langfuse Integration: Provides observability and tracing for agent runs.

Maintenance & Community

Aegra is participating in Hacktoberfest 2025, actively seeking contributions for feature development, bug fixes, and documentation improvements. Detailed contribution guidelines are available, encouraging quality over quantity.

Licensing & Compatibility

Licensed under the Apache 2.0 License, Aegra permits commercial use and integration into closed-source applications without copyleft restrictions, ensuring maximum flexibility for adopters.

Limitations & Caveats

Features such as custom HTTP endpoints, generative user interfaces, and advanced UI theming are listed on the roadmap as upcoming developments, indicating they are not yet available in the current release.

Health Check
Last Commit

3 days ago

Responsiveness

Inactive

Pull Requests (30d)
20
Issues (30d)
14
Star History
105 stars in the last 30 days

Explore Similar Projects

Starred by Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems") and Elvis Saravia Elvis Saravia(Founder of DAIR.AI).

agent-service-toolkit by JoshuaC215

0.5%
4k
Toolkit for building AI agent services
Created 1 year ago
Updated 3 weeks ago
Starred by Elvis Saravia Elvis Saravia(Founder of DAIR.AI), Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and
7 more.

langgraph by langchain-ai

1.3%
21k
Agent orchestration framework for building controllable agents
Created 2 years ago
Updated 18 hours ago
Feedback? Help us improve.