TeamWiseFlow/xiaobei

为OPC/中小微企业量身打造的自媒体获客智能体

小贝（xiaobei） – AI‑驱动的自媒体运营助手

What it is

  • An open‑source AI agent built on the openclaw framework, designed to automate the entire workflow of content creation, publishing, and audience engagement for small‑to‑medium businesses.
  • Interacts through a single WeChat (or Feishu/Enterprise WeChat) chat interface – you type commands and the system handles everything else.

Key capabilities

Category What you can do
Content creation Write, format and push WeChat public‑account articles; generate posts for 小红书/小绿书, Instagram‑style image‑text, and create posters.
Video Produce short videos with B‑roll, slide decks, animated paper‑cut, screen‑recordings; add a digital avatar; supports TikTok, 抖音, 小红书, 快手, YouTube, etc.
Multi‑platform publishing One‑click distribution to WeChat, 视频号, 抖音, 小红书, 快手, TikTok, Twitter/X, 微博, 知乎, Reddit and more.
Growth & analytics Analyze viral videos, replicate top‑account “DNA”, monitor published content, daily recap, comment‑section lead capture.
Customer & market intelligence 24/7 AI客服, market research, competitor intel across 18+ sources (social, news, academic, GitHub) – no API keys required.
Business support Generate product decks, PPTs, IR material, website design, ICP filing assistance, strategic “四声分析”.

How it works

  • The core engine (openclaw) runs a set of crews (agents) such as content‑producer, it‑engineer, and sales‑cs. Crews can call sub‑agents autonomously to solve problems (e.g., refresh an expired token) without user involvement.
  • Skills are shared modules (e.g., smart‑search, browser‑guide) that perform web automation via a patched camoufox anti‑detect browser.
  • Large‑model services are provided mainly through Alibaba Cloud Bailei “Token Plan” (covers chat, image, TTS, ASR, video). Alternate providers (Volcengine, MiniMax) are supported via configuration.

Installation (quick start)

  1. Obtain an Alibaba Bailei Token Plan key (Lite ¥39 / month or Standard ¥139 / month). This single key powers all AI models.
  2. Run the one‑line installer for your OS:
    # macOS / Linux (GitHub)
bash -c "$(curl -fsSL https://raw.githubusercontent.com/TeamWiseFlow/xiaobei/master/scripts/install.sh)"
# Windows PowerShell
irm https://raw.githubusercontent.com/TeamWiseFlow/xiaobei/master/scripts/install.ps1 | iex
    – The script downloads a ~140 MB tarball containing a portable Node + pnpm runtime, the openclaw engine, crew templates, and the camoufox binary.
  3. After installation a QR code appears; scan it with WeChat to bind the bot.
  4. Use the chat window to issue commands like 写一篇关于AI的公众号文章 or 生成抖音短视频并发布.

Running & updating

  • The service runs as an openclaw gateway daemon (systemd/launchd). For testing, use ~/xiaobei/bin/openclaw gateway run.
  • Re‑run the install script to upgrade; it refreshes only the program directory, preserving ~/.openclaw user data.
  • Windows requires Git Bash or WSL for the bash‑based setup scripts.

Configuration highlights

  • Model selection is driven by environment variables (AWK_API_KEY, WORKSPACE_ID, VOLC_SEEDANCE_API_KEY, etc.).
  • Memory can be enhanced by plugging any OpenAI‑compatible embedding model.
  • Optional “dream” nightly memory consolidation can be toggled in plugins.entries.memory-core.config.dreaming.enabled.

Business & support options

  • Free open‑source version plus optional paid VIP Club (¥168 / year) that unlocks premium docs, a WeChat support group, and a free IP‑relay service for platforms that require fixed IPs.
  • Remote installation and custom development services are offered by the Wiseflow team.

License

  • The project is released under the license found in LICENSE (updated from version 4.2 onward).

Bottom line Xiaobei is a full‑stack AI assistant that lets a non‑technical user run a professional content‑creation and distribution operation entirely from a chat app, leveraging large‑model APIs, browser automation, and a modular crew‑based architecture.

Related

  • Project
    ZJU-REAL/Easel

    Easel is an open‑source AI‑driven content workbench for social‑media creators. It links an OpenClaw LLM agent with account‑specific profiles and a library of 113+ “skills” that can discover trends, plan calendars, generate text, images, audio, video, and publish directly to seven Chinese platforms. The project provides a FastAPI + React UI, a CLI installer, and Apache‑2.0 licensing.

  • Project
    ZJU-OmniAI/Data-Copilot

    Data‑Copilot is an open‑source LLM‑based assistant that automatically turns Chinese‑financial‑data queries into full analysis pipelines—fetching data via Tushare, planning tool calls, running predictions, visualising results, and summarising them—all orchestrated by GPT‑3.5 or Qwen‑72B‑Chat. It ships a command‑line entry point and a Gradio web UI, and is released under the MIT license.

  • Project
    itwanger/PaiAgent

    PaiAgent is an open‑source, enterprise‑grade platform that lets users visually compose AI workflows with drag‑and‑drop nodes. It supports multiple large‑model providers via Spring AI, offers two execution engines (a classic DAG engine and a LangGraph4j state‑graph engine), includes a declarative “Skills” system for LLM‑driven function calls, and provides real‑time SSE debugging. The stack is React + TypeScript on the front‑end and Spring Boot 3 + Java 21 on the back‑end, with MySQL (and optional MinIO) for persistence.

  • Project
    tsingyuai/growth-lab

    Growth Lab is an open‑source AI‑agent framework that runs complete product‑growth loops (research, content creation, publishing, measurement) via natural‑language chat. It uses Codex/Claude Code as a runtime, stores context in persistent memory, and ships ready‑made loops for SEO pages, Xiaohongshu content replication, and WeChat article drafting. The repo includes the orchestration code, example Skills, and configuration guidance; data and credentials stay with the user. Licensed under Apache 2.0.

  • Project
    yenche123/liubai

    Liubai is an open‑source, AI‑enhanced personal note‑taking, calendar, and task‑management system. It stores information as modular “cards” that can include text, images, files, tags, and status. The platform works as a responsive web PWA, a WeChat mini‑program, and a VS Code extension, all syncing through a Laf‑based backend. Integrated LLMs (DeepSeek, Kimi, MiniMax, etc.) let you query or augment your notes directly from WeChat, while privacy controls let you keep cards local or block AI access. Offline‑first design, dark/light themes, and Chinese/English UI are built‑in. AGPL‑v3 license.