Skip to content
 
 

Latest commit

 

History

23 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

💫 Ackem

Version Platform License Local First OpenAI Compatible Electron

Ackem · A.C.K.E.M — Autonomous Companion Keeping Emotional Memory
保持情感记忆的自主伙伴

Ackem v1.0.0 — A local-first AI companion for Windows. Bring your own LLM (cloud or local); Ackem handles chat, memory, emotion, relationship state, and desktop presence — all on your hard drive.

Source: GitHub · Gitee mirror
Download: GitHub Releases · Gitee Releases
Build: npm run dist:greendist/release/Ackem-1.0.0-win-x64/ · Path map · Docs languages
Status: Ackem is still in active testing. As a solo-maintained project, test coverage is limited — you may hit unexpected behavior or rough edges. Frequent crashes and severe lag are uncommon, but imperfections still happen. Open an Issue if something feels off.

中文文档 · Privacy & data (EN)


At a glance

Ackem is not a web chat box — it is a Windows desktop app that stays with you: tray, optional desktop pet, structured memory, and a companion that remembers context over time.

💬 Chat Any OpenAI-compatible API — DeepSeek, OpenAI, Ollama, LM Studio, etc.
🧠 Memory Conversations become searchable memory; import .txt / .md as long-term context
💞 Companion Trust, mood, relationship stage, personality presets, diary, optional proactive messages
🔒 Local-first Personal data lives in ./data/ next to the exe — not bundled in the release zip

You need: Windows 10/11 64-bit · an LLM API key or local server · ~10–30 s on first launch (embedding model extracts once). No Node.js required for the green release.


Screenshots & demo

📷 Screenshots (click to expand)

Loading screen
Loading — first launch extracts the local embedding model; main UI opens when the bar completes

Home
Home — chat, memory, games, extensions, settings; relationship & pet preview on the right

Memory visualization
Memory — structured recall, timelines, and knowledge connections from your conversations

Settings
Settings — personality, voice, desktop pet, WeChat bridge, extensions, data

Model & API
Model & API — Base URL, API Key, Model ID

Compliance
Compliance — privacy, data handling, and adult-mode terms on first run

🎬 Demo GIFs (click to expand)

Download and open
Download zip → extract → launch Ackem.exe → wait for loading → main UI

Daily chat
Configure your model, then chat; replies use memory and relationship state


In depth

What is Ackem?

Ackem is a local-first Windows desktop application: you configure your LLM endpoint, and Ackem orchestrates conversation, memory, emotion & relationship state, and desktop companionship while keeping data on your machine.

What you can do

  • Chat like with a person — OpenAI-compatible cloud or local Ollama / LM Studio; configure under Settings → Model & API.
  • Remember what you talk about — structured memory, search, timelines, knowledge graph; import .txt / .md as long-term memory.
  • Continuous companionship — trust, mood, relationship stage, personality presets, companion diary, optional proactive outreach.
  • Beyond the main window — system tray; optional desktop pet (geometric orb + Live2D preview).
  • Optional capabilities — voice STT/TTS, WeChat bridge (phone messages, brain on PC), Extension Center, experimental Plan · OpenForU workspace.
  • Game mode — experimental; play supported games (e.g. Minecraft) with your companion when extensions allow.

Where your data lives

Portable green release stores everything under data/ next to Ackem.exe: chats, memories, diaries, API keys in settings. The official zip does not include data/ — an empty folder is created on first run. No default telemetry to an Ackem server.

Backup, migration, deletion: docs/memory-format.md · docs/distribution-windows.md

For developers

See Developers below, architecture, and the document index.


Quick Start (End Users)

For: downloaded the official Release, no Node.js needed.

Privacy (please read)

Official release does not include After first run, only on your machine
Your memories, chats, imported files data/ (portable, next to exe)
API keys or model credentials Settings in local userData
Maintainer or third-party private data What you configure yourself

Details: docs/distribution-windows.md

Steps

  1. DownloadAckem-v1.0.0-win-x64.zip from GitHub Releases or Gitee Releases
  2. Extract — fully to an SSD path (do not run inside the zip)
  3. LaunchAckem.exe or 启动 Ackem.bat; first launch ~10–30 s (loading screen)
  4. Compliance — accept privacy terms (screenshot)
  5. Configure modelSettings → Model & API: Base URL, API Key (cloud), Model ID
  6. First chat — send a message; optionally import .txt/.md memories

Sharing the zip: never re-pack your personal data/ folder — it contains chats, memory, and keys.

macOS (community build, unofficial) {#macos-community-build-unofficial}

Official support: Windows green release only. The Mac builds below are community-contributed by @deufe — not maintained or fully verified by the core team. See Issue #4.

File Architecture Download
Ackem-1.0.0-mac-arm64.dmg Apple Silicon (M series) v3 DMG
Ackem-1.0.0-mac-x64.dmg Intel Mac v3 DMG

Use v3 only (v1/v2 are outdated). SHA256 and install steps: v1.0.0 Release notes · appendix doc

First launch (Gatekeeper): the build is not Apple-signed. After copying to /Applications, run xattr -cr /Applications/Ackem.app, then right-click → Open once. In-app auto-update is Windows-only.


Developers

Ackem is an Electron app. The renderer depends on window.ackem (preload IPC).
Use npm run dev — do not open the Vite URL in a browser alone.

Prerequisites

  • Windows 10/11 · Node.js 20+ · npm ci

Daily development

cd Ackem-v0.0.0
npm install
npm run dev

Dev data/ is in the repo working tree, separate from green release data/ next to Ackem.exe.

Build & package

npm run build          # Compile → out/
npm run dist:green     # Green release → dist/release/
npm run dist:setup     # Optional NSIS installer

Testing

npm run typecheck
npm test
npm run test:renderer

Seven-System Architecture

# System Description Docs
Overall Electron shell, orchestrator, conversation lifecycle 00-overall-system.md
Brain L0 understanding + L4 memory retrieval & decay 01-brain-system.md
Heart L1 relationship + L2 emotion + L3 expression 02-heart-system.md
Mouth Prompt assembly + LLM calling 03-mouth-system.md
Neural Embedding / vector retrieval 04-neural-system.md
Extension Skill/Plugin/Dispatch/OpenForU 05-extension-system.md
Time Temporal awareness, circadian rhythm, reunion, reflection 06-time-system.md
Data Layer SQLite schema, Repository pattern, migrations 07-data-layer.md
IPC API window.ackem.* preload bridge, push events 08-ipc-api.md

Index: docs/developer/architecture/README.md


Documentation

Languages: English (*.md) · 中文 (*.zh.md) — see I18N.md

Purpose English 中文
Repo paths & build CODEBASE-PATHS.md CODEBASE-PATHS.zh.md
Doc map (maintainers) OPEN-SOURCE-DOC-MAP.md OPEN-SOURCE-DOC-MAP.zh.md
Extension protocol DEVELOPER-EXTENSION-PROTOCOL.md DEVELOPER-EXTENSION-PROTOCOL.zh.md
Developer setup dev-setup.md dev-setup.zh.md
Architecture index architecture/README.md architecture/README.zh.md
Data directory memory-format.md memory-format.zh.md
AI context & retrieval ai-context-and-retrieval-policy.md ai-context-and-retrieval-policy.zh.md
Privacy & data privacy-and-data.md privacy-and-data.zh.md
Local models local-models-windows.md local-models-windows.zh.md
Adult mode & safety adult-and-safety-policy.md adult-and-safety-policy.zh.md
Perception perception-layer.md perception-layer.zh.md
Sensitive capabilities sensitive-capabilities.md sensitive-capabilities.zh.md
Windows distribution distribution-windows.md distribution-windows.zh.md
Indexing & scale indexing-and-scale.md indexing-and-scale.zh.md
Security SECURITY.md SECURITY.zh.md
Contributing CONTRIBUTING.md CONTRIBUTING.zh.md
Code of Conduct CODE_OF_CONDUCT.md CODE_OF_CONDUCT.zh.md

License

This project is open-sourced under the AGPL-3.0 license.

Use case Allowed
Personal learning & research ✅ Yes
Open-source project integration (must remain AGPL-3.0) ✅ Yes
Academic research & citation ✅ Yes
Closed-source commercial product ❌ Commercial license required
SaaS service (source code not provided to users) ❌ Commercial license required
Enterprise private deployment (not open-sourced) ❌ Commercial license required
Closed-source API usage (no modification of source) ⚠️ Gray area, consult us

Commercial Licensing

For commercial use, contact: jasonliu_lyf_2005@qq.com

Contributor Agreement

By submitting a contribution to this project, you agree to the Contributor License Agreement (CLA).

Recognized contributors: CONTRIBUTORS.md

Copyright (C) 2026 Jason Liu (JasonLiu0826)


The open-source edition focuses on local .txt/.md memory and auditable retrieval. For closed-source commercial deployment or SaaS scenarios, see the commercial licensing terms in LICENSE.

About

Ackem — 本地优先 AI 桌面陪伴 · Local-first AI desktop companion. 隐私数据不上传,支持记忆/情绪/扩展/ 。AGPL-3.0

Resources

Code of conduct

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages