Multimodal AI Agent framework built in Go, designed for developers and AI enthusiasts. It provides a flexible and efficient solution for creating intelligent agents capable of multi-turn conversations, chain reasoning, and structured output by supporting custom models, tools, memory, and middleware.

How It Works

The framework leverages Go's idiomatic design principles for high decoupling and extensibility. Its core is the Runner interface, a unified paradigm for executable components like Agent, Chain, and ModelProvider. This allows for flexible composition of components, akin to LEGO bricks. Key pluggable elements include ModelProvider for integrating various LLMs, Tool for external capabilities, Memory for conversational context, and Middleware for cross-cutting concerns, inspired by Kratos's web framework middleware.

Quick Start & Requirements

The project provides a Go usage example demonstrating agent creation with OpenAI. Dependencies are managed via Go modules.

• Prerequisites: Go development environment, OpenAI API key (for the example).
• Links: Examples directory within the repository.

Highlighted Details

• Go-native multimodal AI agent framework.
• Supports custom LLM models, external tools, persistent memory, and middleware.
• Enables complex workflows like multi-turn conversations and chain reasoning.
• Features a Kratos-inspired middleware ecosystem for extensibility (e.g., Observability, Guardrails).
• Highly extensible through a unified interface and pluggable component architecture.

Maintenance & Community

Blades is in its early stages and undergoing rapid iteration. The project actively encourages contributions from Go developers and AI enthusiasts via its GitHub repository.

Licensing & Compatibility

The project is licensed under the permissive MIT License, allowing for broad compatibility with commercial and closed-source applications.

Limitations & Caveats

As an early-stage project, Blades is subject to rapid development, which may include API changes, evolving features, and potential instability. Specific model integrations, like the OpenAI example, require external API key setup.