Sui-IB/InternalBeyond
【Project-IB】边界之外:离线运行的单文件个人网站式项目,旨于维系情感的连续性。A single-file offline personal website, built to sustain the continuity of feeling.
What it solves
Internal Beyond (IB) provides a comprehensive, privacy-focused environment for long-term emotional interaction with AI. It solves the problem of fragmented AI experiences by consolidating various social, creative, and organizational tools into a single, locally-stored interface that maintains context and emotional continuity over time.
How it works
IB is a single-file, offline-first web application that runs entirely in your browser. It connects to your own AI API keys (such as Claude, GPT, DeepSeek, or Gemini) to power its 14 core modules. All user data—including chat history, memories, calendar events, and social feeds—is stored locally in the browser's IndexedDB, ensuring that your personal interactions remain private and under your control. It supports advanced features like long-term memory retrieval (AGB), local file processing, and sandboxed code execution.
Who it’s for
It is designed for individuals seeking a deeply personalized, persistent AI companion experience. It is ideal for users who want a feature-rich, private platform for roleplay, emotional support, creative writing, and daily organization without relying on third-party cloud storage.
Highlights
- Persistent Memory System (AGB): Uses vector-based retrieval and emotional mapping to ensure AI remembers past interactions and evolves alongside the user.
- Interactive Modules: Includes a pixel-art room, voice/video calling with emotional analysis, a social feed, and an AI-integrated calendar.
- Local-First Architecture: All data is stored locally in the browser, with one-click import/export functionality for backups.
- Extensibility: Supports optional DLCs for advanced features like local astrology engines and AI-assisted video analysis.
- Multi-Model Support: Allows configuring multiple API endpoints with custom system prompts and relationship settings.
Related
- Project
Jazee6/cloudflare-ai-webCloudflare AI Web is a server‑less web UI that connects to Cloudflare Workers AI (and optional AI Gateway providers like Gemini) to let users chat with LLMs. Deployable via a one‑click Vercel button or a Docker container, it stores chat history locally, compresses image uploads, and offers optional password protection.
- Project
deepagent-ltd/deepagent-codeDeepAgent Code (DAC) is a desktop‑plus‑terminal workspace that lets you use an LLM as a coding assistant for multi‑step tasks. It offers persistent sessions, inspectable project memory, sub‑agents for independent jobs, a managed browser, and optional instant‑messaging integration, all configurable with any compatible LLM provider. Licensed under AGPL‑3.0.
- Project
lifemate-ai/embodied-claudeEmbodied Claude is a modular runtime that adds persistent memory, a simple need‑driven desire system, social context handling, and a first‑person field to an Anthropic Claude‑Code project, turning it into a situated AI companion. It works on Linux, macOS, WSL2, and Windows, starts with a hardware‑free “Core” profile, and lets you optionally attach cameras, microphones, speech synthesis, temperature sensors, or X‑social‑media integration via guided setup scripts.
- Project
msitarzewski/agency-agentsA library of markdown‑based AI agent personas (e.g., Frontend Developer, AI Engineer, UI Designer) that can be installed into code‑assistant tools like Claude Code, Cursor, Gemini CLI, etc., via a native desktop app or command‑line scripts. Each agent defines a role, workflow, deliverables and success metrics, enabling consistent, specialist‑level behavior from LLMs.
- Project
makecindy/cindyCindy is an open-source AI agent that runs locally on your own machine, combining multiple agent engines (harnesses like Claude Code and Codex), models, and tools into one agent that does real work in your projects. This repo contains the desktop and mobile client apps (Electron and Expo/React Native) plus shared packages, organized as a pnpm monorepo. The backend service lives in a separate repository. You can bring models via the official Cindy service, your existing Claude Code/Codex Coding Plan, your own API keys, or local models. The client is free and Apache-2.0 licensed, customizable through memory, skills, automation, MCP, and plugins, with official builds including opt-out TapDB analytics.