refactor: rebrand and update references to MiniCPM Desk Pet

Author: ZMXJJ

README.md

@@ -1,187 +1,119 @@
-# MiniCPM-Desk-Pet
+<p align="center">
+  <img src="assets/readme%20logo.png" alt="MiniCPM Desk Pet" width="760">
+</p>
 
-[![License: AGPL v3](https://img.shields.io/badge/License-AGPL_v3-blue.svg)](./LICENSE)
-[![Upstream: clawd-on-desk](https://img.shields.io/badge/fork%20of-rullerzhou--afk%2Fclawd--on--desk-orange)](https://github.com/rullerzhou-afk/clawd-on-desk)
-[![Model: MiniCPM5-0.9B](https://img.shields.io/badge/model-MiniCPM5--0.9B-green)](https://github.com/OpenBMB/MiniCPM)
+<p align="center">
+  <a href="./LICENSE"><img src="https://img.shields.io/badge/License-AGPL--3.0--only-blue.svg" alt="License"></a>
+  <a href="https://huggingface.co/openbmb/MiniCPM5-1B-GGUF"><img src="https://img.shields.io/badge/Model-MiniCPM5--1B-green" alt="MiniCPM5-1B"></a>
+  <img src="https://img.shields.io/badge/Platform-macOS%20Apple%20Silicon-lightgrey" alt="Platform">
+</p>
 
-**English** | [中文说明](#中文说明)
+<p align="center">
+  <strong>English</strong> | <a href="README.zh-CN.md">简体中文</a>
+</p>
 
-MiniCPM-Desk-Pet is a local-first desktop pet powered by
-[MiniCPM5-0.9B](https://github.com/OpenBMB/MiniCPM). It combines an
-Electron desktop companion with a lightweight `llama.cpp` inference sidecar,
-so users can chat with MiniCPM locally and let the pet react to coding-agent
-events from Cursor, Claude Code, Codex, and related tools.
+<p align="center">
+  A local-first desktop pet powered by MiniCPM. Download, complete the guided setup, and chat with a tiny companion that lives on your desktop.
+</p>
 
-> **Attribution.** The desktop pet UI in [`clawd-on-desk/`](clawd-on-desk/)
-> is a vendored fork of
-> [rullerzhou-afk/clawd-on-desk](https://github.com/rullerzhou-afk/clawd-on-desk)
-> at upstream commit `5b1f003`. MiniCPM-Desk-Pet adds local MiniCPM5
-> inference, onboarding, GGUF model management, LoRA persona switching, and
-> MiniCPM-powered narration on top of that work. See [`NOTICE.md`](./NOTICE.md)
-> for the full attribution chain.
+---
 
 ## Highlights
 
-- **Local MiniCPM chat**: GGUF inference through `llama-server`, with Metal on
-  Apple Silicon, CUDA on NVIDIA, and CPU fallback.
-- **Desktop pet UI**: an Electron companion that can stay on your desktop and
-  surface chat, task-completion, and agent-status cues.
-- **Coding-agent narration**: Cursor / Claude Code / Codex events can trigger
-  short MiniCPM-powered summaries and reactions.
-- **First-run onboarding**: environment check, accelerator detection, model
-  download, sidecar warmup, and ready state are handled in-app.
-- **Persona LoRA support**: switch or import GGUF LoRA adapters from
-  **Settings -> MiniCPM**; bundled adapters are loaded per request to avoid
-  cross-session contamination.
-- **Small app package**: the app ships a thin gateway and `llama-server`;
-  model weights are downloaded separately on first launch.
-
-## Status
-
-The primary release target is **macOS Apple Silicon (M1+)**. The macOS MVP
-path is working end to end: download the DMG, complete onboarding, download a
-GGUF model, warm up the sidecar, and start chatting with the pet.
-
-Windows and Linux builds share the same sidecar architecture, but they may lag
-behind macOS in QA coverage. The DMG is not notarized yet; see the install
-notes below if macOS Gatekeeper blocks the first launch.
-
-## Install
-
-1. Download the latest macOS Apple Silicon `.dmg` from
-   [Releases](https://github.com/OpenBMB/MiniCPM-Desk-Pet/releases).
-2. Open the DMG and drag the app into `Applications`.
-3. If macOS says the developer cannot be verified, either:
-   - right-click the app icon, choose **Open**, then confirm **Open**; or
-   - run `xattr -cr /Applications/MiniCPM\ Desk\ Pet.app` and open it again.
-4. Follow the five-step onboarding flow:
-   - environment check
-   - accelerator detection
-   - GGUF model download from Hugging Face
-   - sidecar launch and warmup
-   - ready state
-
-No Python, conda, uv, PyTorch, or terminal setup is required for normal use.
-
-## Daily Use
-
-- `Cmd+Shift+M`: toggle the MiniCPM chat bubble.
-- `Cmd+Shift+T`: toggle thinking-mode display.
-- `Esc`: close the bubble when it has focus.
-- Right-click the pet, then open **Settings -> MiniCPM** to change the GGUF
-  path, accelerator, chat-bubble position, narration toggle, onboarding state,
-  or active LoRA persona.
-
-## Persona Adapter
-
-The bundled neko-style LoRA adapter is fine-tuned from MiniCPM5-0.9B using
-[liumindmind/NekoQA-30K](https://huggingface.co/datasets/liumindmind/NekoQA-30K),
-a Hugging Face dataset with 30,834 cat-girl QA samples across categories such
-as ACG, emotional support, creative writing, safety, math, code, and workplace
-scenarios. The Hugging Face dataset card lists the dataset license as
-Apache-2.0.
-
-The adapter is provided as both PEFT source artifacts and a GGUF LoRA file for
-`llama-server`. See [`adapters/README.md`](adapters/README.md) and
-[`adapters/lora_nekoqa_adapter_20260515_0738/USAGE.md`](adapters/lora_nekoqa_adapter_20260515_0738/USAGE.md)
-for conversion details, smoke-test prompts, and known limitations.
-
-## Developer Quickstart
-
-See [`docs/development.md`](docs/development.md) for the full development
-guide. The short path is:
+- **Local by default** — after the model is downloaded, everyday chat runs on your machine.
+- **Zero manual setup** — first launch guides you through environment check, model download, and warm-up.
+- **Desktop companion** — open a floating chat bubble, talk with MiniCPM, and keep the pet on screen while you work.
+- **Agent-aware reactions** — the pet can react to coding activity from tools such as Cursor, Claude Code, and Codex.
+- **Smart model download** — the app can download from Hugging Face or ModelScope and choose the better source for your network.
+- **Persona support** — switch or import character adapters from **Settings -> MiniCPM**.
 
-```bash
-git clone git@github.com:OpenBMB/MiniCPM-Desk-Pet.git
-cd MiniCPM-Desk-Pet
+## Getting Started
+
+### System Requirements
+
+| Item | Recommended |
+| --- | --- |
+| OS | macOS 14.0 or later |
+| Chip | Apple Silicon (M1/M2/M3/M4) |
+| Disk Space | About 2 GB for the app and default model |
+| Network | Required on first launch unless you already have a local model file |
+
+### Installation
 
-mkdir -p models
-# Drop a GGUF you already have, or let onboarding download one on first launch.
-cp /path/to/your/minicpm5-0.9b.Q4_K_M.gguf models/
+1. Go to [Releases](https://github.com/OpenBMB/MiniCPM-Desk-Pet/releases) and download the latest macOS `.dmg`.
+2. Open the DMG and drag **MiniCPM Desk Pet** into `Applications`.
+3. Launch the app and follow the setup guide.
 
-./go.sh        # dev mode: build llama-server if needed, sync gateway, launch app
-./go.sh build  # package a macOS arm64 DMG
+If macOS blocks the first launch, right-click the app and choose **Open**. If needed, remove the quarantine flag:
+
+```bash
+xattr -cr /Applications/MiniCPM\ Desk\ Pet.app
 ```
 
-The inference sidecar lives in [`minicpm-sidecar/`](minicpm-sidecar/) and wraps
-`llama.cpp` with a thin FastAPI gateway. The Electron desktop app lives in
-[`clawd-on-desk/`](clawd-on-desk/).
-
-## Documentation
-
-- [`minicpm-sidecar/README.md`](minicpm-sidecar/README.md): sidecar overview,
-  API, build flow, and vendored `llama.cpp` branch.
-- [`docs/development.md`](docs/development.md): developer setup and repository
-  layout.
-- [`docs/llama-cpp-migration.md`](docs/llama-cpp-migration.md): migration from
-  the older PyTorch sidecar to `llama.cpp`.
-- [`adapters/README.md`](adapters/README.md): LoRA adapter format, conversion,
-  loading, and packaging notes.
-- [`CONTRIBUTING.md`](CONTRIBUTING.md): contribution workflow and test
-  expectations.
-- [`CHANGELOG.md`](CHANGELOG.md): version history.
-- [`docs/archive/`](docs/archive/): archived v0.7 design and architecture docs.
-
-## 中文说明
-
-MiniCPM-Desk-Pet 是一个本地优先的 MiniCPM5-0.9B 桌宠实验项目：Electron 桌宠负责
-交互和事件展示，`minicpm-sidecar/` 通过 `llama.cpp` 的 `llama-server` 加载 GGUF
-模型，在 Apple Silicon 上走 Metal，在 NVIDIA 上走 CUDA，并保留 CPU fallback。
-
-普通用户只需要从
-[Releases](https://github.com/OpenBMB/MiniCPM-Desk-Pet/releases) 下载 macOS DMG，
-双击安装后跟随 Onboarding 完成环境检查、加速器探测、模型下载、sidecar warmup 和
-就绪流程。日常可用 `Cmd+Shift+M` 打开聊天气泡，或在右键菜单的
-**Settings -> MiniCPM** 中切换模型路径、旁白、加速器和 LoRA 人格。
-
-本项目的桌宠 UI 层 fork 自
-[rullerzhou-afk/clawd-on-desk](https://github.com/rullerzhou-afk/clawd-on-desk)
-（upstream commit `5b1f003`），并在此基础上加入 MiniCPM5 本地推理、模型下载、
-Onboarding、LoRA 人格切换和 coding-agent 旁白。完整归属信息请见
-[`NOTICE.md`](./NOTICE.md)。
-
-内置猫娘风格 LoRA 使用 Hugging Face 上的
-[liumindmind/NekoQA-30K](https://huggingface.co/datasets/liumindmind/NekoQA-30K)
-数据集微调，该数据集包含 30,834 条猫娘问答样本，覆盖 ACG、心理疗愈、创意写作、
-安全、数学、代码、职场等类别；Hugging Face 数据集卡片标注其许可证为 Apache-2.0。
+### First Launch
+
+MiniCPM Desk Pet includes a complete first-launch guide:
+
+**Environment Check** -> **Model Download** -> **Model Warm-up** -> **Ready to Use**
+
+The default model is [MiniCPM5-1B-GGUF](https://huggingface.co/openbmb/MiniCPM5-1B-GGUF). You can let the app download it automatically, or choose an existing local `.gguf` file.
+
+## Features
+
+### Chat With a Local Pet
+
+Use the floating chat bubble to talk with MiniCPM from your desktop. Once setup is complete, your normal conversations do not need a remote inference service.
+
+Useful shortcuts:
+
+- `Cmd+Shift+M` — open or close the MiniCPM chat bubble
+- `Cmd+Shift+T` — show or hide thinking mode
+- `Esc` — close the bubble when input is focused
+
+### Reactions While You Work
+
+MiniCPM Desk Pet can stay beside your workspace and react to coding-agent activity: thinking, working, finishing tasks, waiting for attention, or going idle.
+
+### Model Management
+
+The MiniCPM settings page lets you:
+
+- download the default model or choose a local model file
+- rerun onboarding
+- manage character/persona adapters
+- restart the local model runtime when needed
+
+### Persona Adapters
+
+The app includes a neko-style persona adapter. You can switch adapters or import your own from **Settings -> MiniCPM**.
+
+## Roadmap
+
+- Broader Windows and Linux validation.
+- More persona presets.
+- Clearer model download diagnostics and retry guidance.
+- Faster first launch and smaller app footprint.
+- Richer desktop-pet narration for long-running coding sessions.
+
+## Known Limitations
+
+- The primary tested release target is macOS Apple Silicon.
+- First launch requires an internet connection unless you provide a local model file.
+- Response speed depends on your chip, memory pressure, and selected model.
+- Coding-agent reactions depend on each tool's integration behavior and may vary by version.
+
+## Developer Notes
+
+For development setup, packaging, and repository layout, see [`docs/development.md`](docs/development.md).
 
 ## Acknowledgments
 
-- **Desktop pet UI**: forked from
-  [rullerzhou-afk/clawd-on-desk](https://github.com/rullerzhou-afk/clawd-on-desk)
-  at upstream commit `5b1f003`. Thanks to
-  [@rullerzhou-afk](https://github.com/rullerzhou-afk) and all upstream
-  contributors.
-- **Inference backend**: built on
-  [llama.cpp](https://github.com/ggml-org/llama.cpp) (MIT, © Georgi Gerganov
-  and contributors). MiniCPM5 tokenizer support is currently vendored from
-  [zhangtao2-1/llama.cpp@c5ede29](https://github.com/zhangtao2-1/llama.cpp).
-- **Model weights**: loaded from
-  [OpenBMB/MiniCPM](https://github.com/OpenBMB/MiniCPM); weights are not bundled
-  in this repository.
-- **Persona data**: the bundled neko LoRA uses
-  [liumindmind/NekoQA-30K](https://huggingface.co/datasets/liumindmind/NekoQA-30K)
-  for fine-tuning.
-- See [`NOTICE.md`](./NOTICE.md) and
-  [`clawd-on-desk/NOTICE.md`](clawd-on-desk/NOTICE.md) for the authoritative
-  third-party attribution list.
+- Desktop pet UI is based on [rullerzhou-afk/clawd-on-desk](https://github.com/rullerzhou-afk/clawd-on-desk). Full attribution is listed in [`NOTICE.md`](./NOTICE.md).
+- Model weights come from the OpenBMB MiniCPM model family and are downloaded separately.
+- The bundled neko persona uses [liumindmind/NekoQA-30K](https://huggingface.co/datasets/liumindmind/NekoQA-30K) for fine-tuning data.
 
 ## License
 
-This repository is distributed under
-[GNU AGPL-3.0-only](./LICENSE), matching the upstream `clawd-on-desk` license.
-
-- **Source code** in this repository, including `clawd-on-desk/`,
-  `minicpm-sidecar/`, and adapter tooling, is distributed under AGPL-3.0-only.
-- **MiniCPM model weights** are downloaded separately and are governed by the
-  [OpenBMB MiniCPM Model License](https://github.com/OpenBMB/MiniCPM/blob/main/MiniCPM%20Model%20License.md),
-  not by AGPL.
-- **Fine-tuning datasets** keep their own licenses; NekoQA-30K is cited above
-  with its Hugging Face dataset page.
-- **Artwork and theme assets** follow the rights and notices listed in the
-  upstream `clawd-on-desk/NOTICE.md`.
-
-Under AGPL-3.0 section 13, if you run a modified version of this project as a
-network service, you must provide users with access to the complete
-corresponding source code. The public source for this project is
-https://github.com/OpenBMB/MiniCPM-Desk-Pet.
+This repository is distributed under [GNU AGPL-3.0-only](./LICENSE).
+
+MiniCPM model weights are downloaded separately and governed by the [OpenBMB MiniCPM Model License](https://github.com/OpenBMB/MiniCPM/blob/main/MiniCPM%20Model%20License.md). Artwork, third-party code, and datasets keep their own notices; see [`NOTICE.md`](./NOTICE.md) and [`clawd-on-desk/NOTICE.md`](clawd-on-desk/NOTICE.md).