Skip to content

README_EN.md

Latest commit

 

History

History
225 lines (172 loc) · 7.37 KB

README_EN.md

File metadata and controls

225 lines (172 loc) · 7.37 KB

LinX · 灵枢

LinX Logo

License Python FastAPI React Docker Backend Tests Frontend Tests Security Scanning

An intelligent collaboration platform for multi-agent orchestration, project execution, knowledge retrieval, memory management, skill composition, and external runtime integration.

LinX is a full-stack platform that goes beyond a single chat UI. It connects agents, projects, tasks, knowledge, memory, skills, scheduling, and execution environments into one extensible AI collaboration workspace.

What’s in the repo

  • Full-stack implementation with FastAPI + React + TypeScript
  • Multi-module backend for agent orchestration, project execution, skills, knowledge, memory, scheduling, monitoring, and access control
  • Local development stack powered by PostgreSQL, Redis, Milvus, MinIO, and Docker Compose
  • User-facing product surfaces for projects, runs, agent conversations, knowledge management, skills, schedules, settings, and admin operations
  • Optional external Runtime Host support through a one-line installer generated by the control plane

Core capabilities

Agent orchestration

  • Agent creation, configuration, testing, logs, and runtime metrics
  • Persistent conversations with Markdown rendering, attachments, and workspace files
  • Runtime access to skills, knowledge, and user memory
  • Support for external runtimes and archived session workspaces

Project execution

  • End-to-end flows for projects, project tasks, execution plans, runs, and run steps
  • Run Center and Run Detail pages for routing, timelines, deliverables, and external sessions
  • Execution nodes, lease dispatch, and project workspace sync support

Skills and extensions

  • Skill creation, editing, testing, versioning, package import, and bindings
  • Candidate review / merge / reject workflows
  • SKILL.md-style skills plus LangChain-style tool integrations
  • MCP Server management and tool synchronization

Knowledge and memory

  • Document upload, download, reprocessing, preview, search, and retrieval testing
  • Processing pipeline for parsing, chunking, enrichment, and indexing
  • User memory search, configuration, retrieval testing, and detail views

Platform operations

  • Schedules with create, edit, pause, resume, run-now, and detail views
  • Dashboard, notifications, departments, users, roles, settings, and profile management
  • JWT auth, request logging, rate limiting, health checks, and Docker-based sandbox execution

Architecture

graph TD
    UI["React Frontend"] --> API["FastAPI API Gateway"]
    API --> AGENT["Agent Framework"]
    API --> EXEC["Project Execution & Task Manager"]
    API --> SKILL["Skill Library & MCP"]
    API --> KB["Knowledge Base"]
    API --> MEM["User Memory"]
    API --> AUTH["Auth / RBAC / Admin"]
    API --> WS["WebSocket / Notifications / Telemetry"]

    AGENT --> LLM["Ollama / OpenAI / Anthropic / vLLM"]
    EXEC --> NODE["External Runtime / Node Agent"]
    API --> PG[("PostgreSQL")]
    API --> REDIS[("Redis")]
    KB --> MILVUS[("Milvus")]
    API --> MINIO[("MinIO")]
Loading

Detailed architecture: docs/architecture/system-architecture.md

Tech stack

Backend

  • FastAPI, SQLAlchemy, Alembic
  • LangChain plus custom agent framework
  • PostgreSQL, Redis, Milvus, MinIO
  • WebSocket, monitoring, and Docker sandbox integrations

Frontend

  • React 19, TypeScript, Vite
  • Zustand
  • Tailwind CSS v4, Framer Motion
  • Recharts, React Flow, Monaco Editor

Infrastructure

  • Docker Compose
  • Optional Kubernetes
  • Optional external Runtime Hosts via control-plane install scripts

Quick start

Option 1: Docker Compose

Recommended for first-time setup.

Requirements

  • Docker 24+
  • Docker Compose 2.20+
  • Recommended memory: 8 GB+

1) Copy environment variables

cp .env.example .env

Review at least:

  • POSTGRES_PASSWORD
  • REDIS_PASSWORD
  • MINIO_ROOT_PASSWORD
  • JWT_SECRET
  • OLLAMA_BASE_URL or cloud model credentials

2) Start services

docker compose up -d

3) Open the platform

  • Frontend: http://localhost:3000
  • Backend API: http://localhost:8000
  • Swagger docs: http://localhost:8000/docs
  • MinIO Console: http://localhost:9001

If the system is not initialized yet, the frontend will guide you through the first-run setup flow.

Option 2: Local backend/frontend development

Backend

Please use the project virtual environment at backend/.venv.

cd backend
python3 -m venv .venv
.venv/bin/pip install -r requirements.txt
.venv/bin/pip install -r requirements-dev.txt
.venv/bin/python scripts/dev_server.py preflight
make run

Common commands:

cd backend
source .venv/bin/activate
make run
make run-debug
make format
make lint
make type-check
make test
make test-cov
make migrate

Frontend

cd frontend
npm install
npm run dev

Common commands:

cd frontend
npm run lint
npm run type-check
npm run format
npm run test
npm run build

Optional: external runtime host

Open an external agent's Runtime Host panel to generate the one-line install command for a target machine.

Configuration

Project structure

.
├── backend/               # FastAPI backend and platform modules
├── frontend/              # React + TypeScript frontend
├── docs/                  # Architecture, backend, developer, and deployment docs
├── infrastructure/        # Docker / Kubernetes resources
├── docker-compose.yml     # Local / test stack
├── LICENSE                # Dual-license overview (MIT OR Apache-2.0)
├── LICENSE-MIT            # MIT license text
└── LICENSE-APACHE         # Apache License 2.0 text

License

LinX is dual-licensed under MIT or Apache-2.0 at your option. See LICENSE, LICENSE-MIT, and LICENSE-APACHE.

Contributing

Please read:

We’re happy to receive issues, pull requests, and documentation improvements.