Znote is a local-first Markdown editor with AI and code built in. It works directly on the plain .md files on your disk, lets you ask your own notes questions with AI, runs JavaScript code blocks inline with rendered charts and tables, connects to real tools through MCP, and versions your vault with git — all free, offline, and account-free.
| 面向/指標 | ![]() | |
|---|---|---|
| 官方網域 | codegpt.co | znote.io |
| 定價模式 | 免費增值、付費、需聯絡報價 | 免費增值、付費 |
| 每月造訪次數 | 17.01萬 (170,064) | 3761 (3,761) |
| 月流量變化量 | -3.98萬 (-19%) |
| +123 (+3.4%) |
| 主要市場與市佔 | 🇮🇳 印度 (13.6%) | 🇺🇸 美國 (94.1%) |
|---|
| 每月直接造訪 | 9.21萬 | 1468 |
|---|
| 使用者評分 | 4.3 | 4.4 |
|---|
| 支援平台 | 網站 | 網站 |
|---|

支援 VS Code、JetBrains 與 Visual Studio 的 AI 程式智能代理,自帶金鑰自由選模型
以自備 API 金鑰串接 Claude、GPT、Gemini、Grok 或本機 Ollama/LM Studio 模型,也可使用 CodeGPT 額度,並依任務切換模型且無需搬移程式碼。
智能代理先讀取程式碼庫、提出可編輯方案,只在你逐步確認後才寫程式,讓每處改動都在掌控之中。
可查看讀取了哪些檔案、權衡了哪些方案、提出了哪些修改,並即時追蹤上下文用量,沒有隱藏上限。
透過 MCP 串接外部 API、文件與資料庫,並設定專案規則,讓生成的程式碼符合團隊規範。
Znote is a local-first Markdown editor with AI and code built in. It works directly on the plain .md files on your disk, lets you ask your own notes questions with AI, runs JavaScript code blocks inline with rendered charts and tables, connects to real tools through MCP, and versions your vault with git — all free, offline, and account-free.
Type @ to choose which of your notes the AI reads and / to pick a prompt; the answer lands inline in the note you are writing, with no chat window or copy-paste.
Write a JavaScript block, press play, and query a database, call an API, or parse a file — the chart or table renders inline and the saved note becomes a reusable tool.
Connect real tools through MCP servers such as GitHub, Postgres, Slack, and Linear, so AI blocks can query, search, and create on your behalf using the same config format as Claude Desktop.
Your vault is plain, git-compatible files. Pull, commit, and push from the sidebar, while note passwords stay encrypted in a sidecar file so secrets never reach a commit.
A comprehensive agentic development suite by Google for running parallel coding agents, but requires a separate environment.
#2
#3A spec-driven AI IDE from AWS that plans and verifies engineering work, but requires switching editors.

專項更強Optimized for running open-weight models in the terminal, focusing on personal coding preference adaptation.

專項更強自動化 Pull Request 的深度程式碼審查,而 CodeGPT 則主要專注於編輯器內的積極生成與重構。
A comprehensive agentic development suite by Google for running parallel coding agents, but requires a separate environment.
專項更強Optimized for running open-weight models in the terminal, focusing on personal coding preference adaptation.