Weave

Weave is a high-performance Golang-based MCP server that implements a microkernel + layered architecture, focusing on flexible plugin management and AI integrations such as LLM, AIChat, Agent, and RAG. The server exposes a unified plugin interface, allowing developers to drop in plugins that register routes under their own namespaces and manage lifecycle, dependencies, and conflicts at runtime. With Weave, you can rapidly assemble AI-enabled applications by combining tools, data services, and automation workflows into a cohesive, scalable platform.

To use Weave, run the container (or install from source if you have the Go environment) and interact with its plugin-based API surface. Plugins can be loaded hot-plug style, so you can add new capabilities without restarting the core kernel. The platform is designed to handle API gateway routing, service orchestration, and AI-driven workflows, enabling you to build LLM-powered apps, AIChat experiences, RAG pipelines, and agent-driven automation in a modular fashion.

Prerequisites:

• Docker installed and running
• Basic knowledge of container workflows
• Optional: Go development environment if you plan to build or customize plugins

Install via Docker (recommended):

1. Pull and run the Weave image: docker pull liaotxcn/weave docker run -d --name weave -p 8080:8080 liaotxcn/weave

Install from source (Go-based development path):

1. Install Go (make sure you have Go 1.20+ in your environment).
2. Clone the repository: git clone https://github.com/user-attachments/weave.git cd weave
3. Build the binary: go build ./...
4. Run the server: ./weave-server

Configuration:

• Weave uses a plugin-based configuration. You can extend behavior by adding plugins in the plugins/ directory or wiring them via the core kernel configuration.
• Environment variables can override core settings (e.g., configuration paths, logging levels, security).

Tips and common considerations:

• Weave supports hot-plug plugins, so you can enable/disable features without restarting the core kernel.
• Each plugin can register its own route namespace. Use consistent naming to avoid route conflicts.
• When deploying in production, consider mounting a configuration store (e.g., Consul, Etcd) for dynamic configuration; enable secure defaults for logs and secrets.
• If you encounter routing or lifecycle issues, check plugin dependency resolution and conflict detection messages in the kernel logs.
• For AI integrations, ensure that LLM/AI capabilities have appropriate API credentials and quota configured in environment variables or secrets management.