TreeMap
A free, open-source disk-space visualizer. Squarified treemap, duplicate hunting, disk-full forecasting, and cleanup that only ever moves files to the Trash.
翻譯準備中
個人 GitHub 帳號下公開儲存庫的擁有者可以登入並驗證所有權,自行管理介紹,無需管理員核准。
登入並產生驗證檔案,在 30 分鐘內將其提交至此儲存庫的預設分支。其他分支或分叉儲存庫中的檔案無效。然後在下方檢查檔案,即可無需管理員核准直接管理專案。請保留檔案以維持管理權限。
如需登入,將前往登入頁面。登入後請回到此處,再次點擊所選的驗證按鈕。
在專案資料夾中執行 /nomorevibe verify,即可驗證所有權並自行管理介紹。
一分鐘完成驗證One Desktop Agent. Infinite Personal Tools.
翻譯準備中
取自儲存庫 README
正在載入完整 README。
檢視完整 README ↗Kucedr is a desktop AI assistant that brings conversation, tools, and your personal context into one workspace. Ask it to summarize a document, research a question, edit project files, or create images, audio, and video. Type or speak, attach files, and follow the agent's tool activity as it works through the task.
Choose providers and models separately for chat, voice, and media. Shape the assistant with instructions, memory, and a searchable knowledge base; connect services through MCP; and schedule recurring tasks and health checks while Kucedr is running. Extend your workspace with focused apps, reusable skills, and compatible remote agents.
Conversations, settings, memory, and workspace files live in your local Kucedr profile. A Kucedr account is optional for local use. Tasks that use external models or connected services send the required inputs to those services; selected folders can also be backed up or synchronized remotely. See Control and Privacy for credential handling and data boundaries.
Start with the application guide, or explore the full documentation index.
Kucedr runs on Windows, macOS, and Linux, with English and Italian interfaces and light, dark, and system themes.
Choose a package for your operating system from GitHub Releases. After launch, configure a model provider and select an assistant model. You can continue local-only without an account; provider usage may require your own credentials and incur charges.
Requirements: Node.js 22.19+ and npm 11.5.1+.
npm ci
npm run dev
The root install includes the Electron app, @kucedr/sdk, and @kucedr/cli through npm
workspaces and one lockfile. The manifest also declares a website workspace, but that
directory is absent in this checkout; website scripts require it to be supplied separately.
On first launch, follow the Start Page Flow to sign in or continue local-only, save a model-provider API key, and select the provider and model for the assistant. Search, database, speech, and media configuration are optional and can be completed later in Settings. Configure an S3-compatible provider and select folders for backup in Settings → Storage. Folder backup does not require account sign-in. Version history sync requires sign-in and a configured backend; see Cloud file synchronization. See Home UI for the chat workspace's states and interactions. See Settings UI for configuration navigation and behavior.
If the normal development command cannot start on your Linux host, the following local-development command disables Electron's sandbox. Do not use this override for production distribution:
npm run dev-linux
The TypeScript CLI lives in packages/cli. It launches the desktop app, installs validated Kucedr
plugins, and includes an interactive terminal interface:
npm run cli:build
npm link ./packages/cli
kucedr
kucedr install package-one
kucedr tui
The CLI validates and stores plugin packages, but the desktop runtime does not automatically
activate contributions from the CLI install directory. See Plugins for supported
runtime paths. Inside the TUI, enter /install package-one. See
packages/cli/README.md for the command and plugin-install contracts.
For documentation-only changes, check the touched Markdown with Prettier and verify relative links. See Contributing for the workflow.
For code changes, run the main local checks before submitting:
npm run quality:check
This runs dependency audits, TypeScript checks, ESLint, main-process tests, renderer tests, and package tests. Run the end-to-end suite separately:
npm run test:e2e
npm run build # Type-check and create a production build
npm run dist:win # Windows x64 installer and portable executable
npm run dist:win:portable # Windows x64 portable executable only
npm run dist:mac # macOS package for x64 and arm64
npm run dist:mac:dmg # macOS DMG for x64 and arm64
npm run dist:linux:appimage # Linux AppImage
npm run dist:linux:portable # Linux AppImage and tar.gz archive
With Node.js 22 or newer, run bash scripts/install.sh on macOS or Linux, or from Git Bash on Windows.
It selects the latest published Kucedr release for the current OS and architecture, then places
the portable app in %LOCALAPPDATA%\\Programs\\Kucedr, ~/Applications, or ~/.local/bin,
respectively. Windows and Linux builds support x64; macOS builds support x64 and arm64.
If no matching Kucedr artifact has been published, the script reports that instead of installing
an older Friday release or an installer package.
On Windows, download Kucedr-Portable-<version>-x64.exe and run it directly. It temporarily
extracts its application files while Kucedr is running, but does not install shortcuts,
file associations, or uninstall records and does not require administrator access.
On Linux, download the AppImage, mark it executable, and launch it. If AppImage mounting or FUSE
is unavailable, extract the .tar.gz release and run kucedr-desktop from the extracted folder.
Neither option requires a package installation.
Kucedr settings, conversations, workspace files, and generated data remain under
%USERPROFILE%\.kucedr on Windows or $HOME/.kucedr on Linux. Electron runtime data remains in
%APPDATA%\Kucedr on Windows or $XDG_CONFIG_HOME/Kucedr on Linux, normally
$HOME/.config/Kucedr. Portable updates are manual: close Kucedr and replace the executable or
extracted application; the profile data remains in place.
Portable packaging does not bypass AppLocker, WDAC, Linux noexec, endpoint security, or network
policy. Protected command execution may require administrator or IT setup, and browser automation
requires an installed, permitted Google Chrome. Kucedr reports these limitations without preventing
chat and other supported features from running.
src/main contains the Electron main process, agent runtime, channels, model integrations, media services, transcription, IPC, and application services.src/renderer/src contains the React user interface.src/preload exposes the narrow bridge between the renderer and main process.src/shared contains cross-process types and API contracts.src/main/terminal contains the PTY lifecycle behind the typed terminal IPC API. See Terminal IPC Architecture.packages/cli contains the publishable TypeScript command-line and terminal interface.packages/sdk contains typed embedded bridges and an HTTP client; this checkout does not start
a matching standalone SDK HTTP server. See Application Architecture.src/main/agent contains sessions, tools, skills, memory, schedules, health runs, sandboxing, and permission policy.src/main/models contains provider-specific model integrations. See
Provider Reference for the built-in catalog and runtime support matrix.src/main/cloud and src/main/storage keep account and cloud behavior behind replaceable ports.
See Account and Cloud Architecture.Renderer windows use sandboxing, context isolation, disabled Node integration, and web security. Preload APIs expose narrow typed IPC methods, and agent writes, edits, patches, and command execution are subject to the permission policy.
See SECURITY.md for the security policy and vulnerability reporting process.
The Electron app, SDK, and CLI are versioned independently. Checked-in automated release workflows are currently disabled; follow the local release checks before publishing. See Development, Testing, and Deployment for local setup, test commands, normal pushes, tag conventions, npm trusted publishing, desktop signing, and recovery procedures.
See CONTRIBUTING.md for setup, workflow, and code standards. Behavioral guidelines for AI-assisted contributions are in AGENTS.md.
MIT © 2026 Harald Bregu