zgi  by zgiai

Platform for building and operating AI agents and workflows

Created 1 year ago
302 stars

Top 88.1% on SourcePulse

GitHubView on GitHub
Project Summary

ZGI provides a self-hostable platform for building and operating AI agents designed for real-world tasks beyond simple chat interactions. Targeting development teams, it enables the creation of sophisticated AI applications by integrating agent orchestration, visual workflow design, dataset management, and model integration, facilitating a swift transition from prototype to production.

How It Works

ZGI functions as an Agent Runtime, unifying several key components within a single workspace. It supports building agent applications with defined instructions, memory, and skills; designing multi-step workflow automations involving LLM calls, code execution, and approvals; and developing reusable runtime skills for tasks like file generation, chart creation, and database access. Agents can be bound to specific knowledge bases and databases, enhancing security and control. The platform also includes model routing for managing providers, credentials, and policies, all within a self-hostable runtime environment comprising console, API, sandbox, runner, PostgreSQL, and Redis.

Quick Start & Requirements

To start the full local stack, run: make dev-docker Alternatively, execute the startup script directly if make is unavailable: ./dev/start-docker Access the console at: http://localhost:2679. An administrator account must be created on first launch. Development requires Docker, Docker Compose, Go, Node.js, and pnpm. Dependencies can be prepared with make setup.

Highlighted Details

  • Builds agent applications with instructions, model settings, memory, knowledge, file upload, and skills.
  • Enables workflow automation with LLM calls, branches, loops, approvals, code execution, and retrieval.
  • Provides reusable skills for file operations, charts, reports, scheduling, calculations, databases, and workflow calls.
  • Supports binding agents to approved knowledge bases and database tables.
  • Includes model routing for managing providers, defaults, credentials, policies, and pricing.
  • Offers a self-hosted runtime environment for console, API, sandbox, runner, PostgreSQL, and Redis.

Maintenance & Community

The repository includes links to CONTRIBUTING.md, CODE_OF_CONDUCT.md, and SECURITY.md for guidance on contributions and security. Specific details regarding active community channels, notable contributors, or project roadmap are not detailed in the provided README.

Licensing & Compatibility

ZGI is available under the ZGI Community License, based on Apache License 2.0 with additional conditions. It is free for personal, research, educational, and internal organizational use. Hosted multi-tenant services, white-label distribution, or removal of ZGI branding require a commercial license. This license is not an OSI-approved open-source license.

Limitations & Caveats

The ZGI Community License imposes restrictions, requiring a commercial license for specific deployment scenarios like multi-tenant hosting or white-labeling, and it is not an OSI-approved open-source license. An administrator account must be manually created upon initial setup.

Health Check
Last Commit

1 week ago

Responsiveness

Inactive

Pull Requests (30d)
0
Issues (30d)
2
Star History
279 stars in the last 30 days

Explore Similar Projects

Starred by Elie Bursztein Elie Bursztein(Cybersecurity Lead at Google DeepMind), Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and
7 more.

SuperAGI by TransformerOptimus

0.0%
18k
Open-source framework for autonomous AI agent development
Created 3 years ago
Updated 1 year ago
Feedback? Help us improve.