ai-chatbot-framework
A lean, self-hosted AI agent that everyone can use. Each person gets an assistant with its own private computer, their files, skills, app connections and schedu
What is ai-chatbot-framework?
A lean, self-hosted AI agent that everyone can use. Each person gets an assistant with its own private computer, their files, skills, app connections and scheduled jobs.
Overview of ai-chatbot-framework
What it is
AI Chatbot Framework is a lean, self-hosted AI agent. Each user gets an assistant with its own private computer, files, skills, app connections, and scheduled jobs. Version 2 rebuilds the original intent/entity NLU chatbot builder as an agent.
Key features
- Any model through LangChain – Claude (default), OpenAI, any OpenAI-compatible server (Ollama, OpenRouter, vLLM, LiteLLM) or any other LangChain chat provider.
- A lean, modern agent loop with four tools (
read,write,edit,bash) plusmcp_callfor MCP tools; parallel calls, background commands, context compaction, steering (type while the agent works), and Stop at any time. - Sub-agents that work side by side, each with its own goal, fresh context, and a model the agent picks.
- One private Docker sandbox per user with persistent
/workspace, resource limits, and no Linux capabilities. - Files, not forms –
context.mdandmemory.mdare plain files the agent reads and edits itself; the UI is a read-only file browser with preview, upload, and download. - Learns as it goes – a background pass updates
memory.mdand skills every few messages; past chats are saved as Markdown for searching. - Skills in the open Agent Skills format, MCP support with OAuth sign-in, personal apps (up to 5 small apps in the menu), views in chat (charts, plans, forms), projects, scheduled tasks, and Agent Client Protocol (ACP) over WebSocket.
Use cases
Personal AI assistant with its own files and memory; automated research and report generation; task scheduling and proactive reminders; building small personal apps without a backend; multi-model comparison via the council-of-llms skill; and integrating external tools through MCP servers.
Getting started
shgit clone https://github.com/alfredfrancis/ai-chatbot-framework.git && cd ai-chatbot-frameworkcp .env.example .env # set ACF_DOMAIN, POSTGRES_PASSWORD, ACF_SECRET_KEY and a model key# prepare the host first: gVisor, user-namespace remapping, sandbox firewall (deploy/README.md)docker compose up -d # or: make up
License and pricing
License: MIT. Pricing: open source.
這是什麼
AI Chatbot Framework 是一個精簡、自託管的 AI 代理。每位使用者都擁有自己的助理,配備專屬的私人電腦、檔案、技能、應用程式連線和排程任務。第 2 版將原本的意圖/實體 NLU 聊天機器人建構器重新打造為代理。
主要功能
- 透過 LangChain 支援任何模型 – Claude(預設)、OpenAI、任何相容 OpenAI 的伺服器(Ollama、OpenRouter、vLLM、LiteLLM)或任何其他 LangChain 聊天提供者。
- 精簡的現代代理循環,包含四個工具(
read、write、edit、bash)加上用於 MCP 工具的mcp_call;支援平行呼叫、背景命令、上下文壓縮、引導(在代理工作時輸入文字),以及隨時停止。 - 子代理可並行工作,各自擁有自己的目標、全新的上下文以及代理選擇的模型。
- 每位使用者擁有專屬的 Docker 沙箱,包含持久的
/workspace、資源限制,且無 Linux 權限。 - 以檔案為核心,而非表單 –
context.md和memory.md是純文字檔案,由代理自行讀取和編輯;使用者介面是唯讀的檔案瀏覽器,支援預覽、上傳和下載。 - 邊用邊學 – 每隔幾則訊息,背景程序會更新
memory.md和技能;過去的對話會儲存為 Markdown 格式以便搜尋。 - 採用開放代理技能格式的技能、支援 OAuth 登入的 MCP、個人應用程式(最多 5 個小型應用程式顯示在選單中)、聊天中的檢視(圖表、計畫、表單)、專案、排程任務,以及透過 WebSocket 的代理用戶端協定(ACP)。
使用情境
擁有個人檔案和記憶的個人 AI 助理;自動化研究與報告生成;任務排程與主動提醒;無需後端即可建構小型個人應用程式;透過 council-of-llms 技能進行多模型比較;以及透過 MCP 伺服器整合外部工具。
快速開始
shgit clone https://github.com/alfredfrancis/ai-chatbot-framework.git && cd ai-chatbot-frameworkcp .env.example .env # 設定 ACF_DOMAIN、POSTGRES_PASSWORD、ACF_SECRET_KEY 和一個模型金鑰# 先準備主機:gVisor、使用者命名空間重新對應、沙箱防火牆(deploy/README.md)docker compose up -d # 或:make up
授權與價格
授權:MIT。價格:開源。
Based on the project's public README and repository metadata.