Skillget

Saki-AI

mcp
Guvenlik Denetimi
Basarisiz
Health Gecti
  • License — License: Apache-2.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 11 GitHub stars
Code Basarisiz
  • network request — Outbound network request in backend/package.json
  • network request — Outbound network request in backend/qqbot/src/api.ts
  • process.env — Environment variable access in backend/qqbot/src/channel.ts
  • process.env — Environment variable access in backend/qqbot/src/config.ts
  • fs.rmSync — Destructive file system operation in backend/qqbot/src/image-server.ts
  • network request — Outbound network request in backend/qqbot/src/image-server.ts
  • process.env — Environment variable access in backend/qqbot/src/onboarding.ts
Permissions Gecti
  • Permissions — No dangerous permissions requested

Bu listing icin henuz AI raporu yok.

SUMMARY

A creative AI agent that lives on your PC. Codes, researches, builds slides, remembers you, and turns your stories into cocktails. Any model, local or cloud. Breaks something? Roll it back in one click.

README.md

Saki AI Agent

Local-first AI workspace for code, files, research and long-term collaboration

Local-first AI Workspace · Model Choice · Recoverable Workflows · Beyond Coding

English · 简体中文

Node.js
React
Vite
Tailwind CSS
Express
License
PRs Welcome

Saki AI is a local-first AI workspace built for Chinese-speaking users. Connect local or cloud models in one GUI, let the agent work on your code, workspace files and terminal tasks, and bring research, documents, slides, long-term memory and QQ workflows into your daily work.

Saki takes the task-driven collaboration of coding agents like Codex and Claude Code as its reference point. It does not claim to out-code them. The goal is a personal AI workspace that isn't locked to one model, covers more than coding, and lets you manage and roll back tasks on your own machine. The UI is served by a local web service; the Windows portable build launches it from an EXE and opens your browser. It is not a native desktop GUI.

Download Windows Portable · Quick Start · Low-resource Deployment · Plugins

Keywords: Desktop AI Agent · Coding Agent · Local-first · Ollama / LM Studio · MCP · Workspace Checkpoints · Long-term Memory · Research & Office

https://github.com/user-attachments/assets/91ac7121-2a01-4751-a0c3-bc257e097b0e

A closer look at Saki

Your ideas, models and tools, together in a personal workspace.

Saki AI workspace — a local-first home for ideas, multi-model collaboration and long-term memory

Extend your workflow with plugins, MCP tools and dedicated Skills.

Saki AI plugin marketplace — discover plugins, MCP tools and dedicated Skills

Start on the desktop, keep exploring

Add new abilities to your AI workflow and take it all the way to your servers. These two projects show what else Saki can do.

Extensions — explore the saki-plugins repository

Find plugins, MCP tools and Skills for everyday tasks. Saki's plugin marketplace installs plugins straight from the repository, or you can start from an example and build your own.

Browse extensions · View repo ↗

 Saki Panel — open-source server panel with an AI assistant

From your desktop to your servers. Manage Minecraft, Steam, Docker and regular processes in one web console, and let Saki troubleshoot with logs, files and node metrics.

Meet Saki Panel · View repo ↗

What it does

  • Any model: Ollama, LM Studio, OpenAI, Anthropic, Gemini, DeepSeek, Zhipu, Kimi, Qwen, Doubao, MiniMax, or any OpenAI-compatible endpoint. Keys are stored per provider; switch models from the top bar.
  • Real work in your repo: reads and edits files, runs terminal commands, and shows every tool call and diff.
  • Undo anything in the workspace: each turn snapshots the workspace before the agent starts, including changes made from the terminal. Jump back to any turn. No Git required.
  • Steer while it runs: add instructions mid-task, queue follow-ups, edit or cancel them.
  • Long-term memory: preferences and project rules carry across sessions.
  • Beyond code: PDF / Office Q&A, multi-source deep research, PPT generation and export, claim checking.
  • Story Glass: tell Saki a story and she mixes a cocktail for it, served as a shareable card.
  • Extensible: MCP, Skills, and a plugin marketplace. Optional QQ bot, Stable Diffusion and GPT-SoVITS voice.
  • Lightweight: one process, one port in production. Data lives in local JSON files, no database needed.

Quick Start

Windows portable: download Saki-AI-3.0-win-x64.zip, unzip, run Saki-AI.exe. Your browser opens http://127.0.0.1:5480. Add a model in Settings and you're set. Node.js is bundled.

From source (Windows / macOS / Linux, Node.js 22.12+):

git clone https://github.com/EthanChan050430/Saki-AI.git
cd Saki-AI
node scripts/launch.cjs

The first launch installs dependencies and builds the frontend. On macOS / Linux, bash deploy.sh start runs it in the background. For low-spec machines, see the deployment guide.

Heads up

Approvals and path checks are app-level guards, not an OS sandbox. Terminal commands and third-party plugins run with your user's permissions, so only grant the folders you need. Checkpoints restore workspace files only; requests and messages already sent can't be undone.

Contributing

Issues and PRs welcome. Licensed under Apache 2.0.

Yorumlar (0)

Sonuc bulunamadi