riverpod
リアクティブなキャッシュおよびデータバインディングフレームワーク。Riverpod は非同期コードの扱いを簡単にする。
Flutter開発者向けに設計された、Claude用のエージェント型ターミナルUI
リアクティブなキャッシュおよびデータバインディングフレームワーク。Riverpod は非同期コードの扱いを簡単にする。
複数のパッケージ(モノレポ)を含むDartおよびFlutterリポジトリを管理するためのツール。コンベンショナル・コミットによる自動バージョニングをサポート。JavaScriptのLernaパッケージにインスパイアされています。
Dart用のクラッシュレポートライブラリで、クラッシュレポートをSentry.ioに送信します。このライブラリはDart VMとWebをサポートしています。Flutterを使用する場合は、代わりにsentry_flutterをご利用ください。
コンポーネント、状態管理、アニメーションなどを使用して、豊かなターミナルユーザーインターフェースを構築する、Flutterに似たフレームワーク。
GNU および POSIX スタイルのオプションを使用して、生のコマンドライン引数をオプションと値のセットにパースするためのパーサーを定義するライブラリ。
{"git":{"url":"https://github.com/Norbert515/nocterm.git","ref":"v0.5.1"}}{"git":{"url":"https://github.com/Norbert515/nocterm.git","ref":"v0.5.1","path":"packages/nocterm_riverpod"}}any^2.4.0anyanyanyanyany^2.5.1^2.3.5^2.4.1^4.9.0^7.3.0^3.1.2^0.9.0^0.2.0^4.5.2^8.12.0^1.2.0^2.4.9^2.5.2^6.8.0^7.3.0^2.4.0^1.26.3以下は英語原文のスナップショットです。最新版は GitHub をご覧ください。
Vide
The open source agent orchestrator. Instead of one AI conversation, Vide orchestrates a team of specialized agents that research, implement, review, and test your code -- all working in parallel.
# macOS / Linux
curl -fsSL https://raw.githubusercontent.com/Norbert515/vide_cli/main/scripts/install.sh | bash
# Windows
irm https://raw.githubusercontent.com/Norbert515/vide_cli/main/scripts/install.ps1 | iex
Requires Claude Code CLI (npm install -g @anthropic-ai/claude-code).
Then just run:
vide
Most AI coding tools give you one agent in one conversation. Vide gives you a team.
A lead agent breaks down your request and spawns specialists -- researcher, implementer, tester -- that work in parallel across separate git worktrees. They communicate asynchronously and iterate through review rounds until the job is done.
Vide agent team in action
Agents don't just run in parallel -- they collaborate. The orchestrator delegates to an implementer, a QA agent reviews the result, issues get fixed, another round of review. This loop repeats until quality is met.
Structured engineering workflows, not just autocomplete.
Run your agent teams on a remote server and connect from anywhere. Vide includes a daemon that manages sessions as sub-processes, with a REST API and WebSocket streaming for real-time control.
# Start the daemon
vide daemon start
Then open Vide and press Tab to open settings -- enable daemon mode there. The TUI works identically in daemon mode -- the same UI, same workflows, just backed by daemon-managed sessions that persist even if the TUI disconnects.
Run the daemon on a VPS and connect from your laptop:
# On the VPS
vide daemon start --host $(tailscale ip -4) --port 8080
# From your laptop
vide connect 100.64.12.34:8080
Tailscale is the recommended way to handle remote access. It creates encrypted peer-to-peer tunnels between your devices using WireGuard -- no port forwarding, no public exposure, no extra auth layer needed.
1. Install on both your VPS and laptop:
# VPS (Ubuntu/Debian)
curl -fsSL https://tailscale.com/install.sh | sh
sudo tailscale up
sudo systemctl enable --now tailscaled
# macOS laptop
brew install tailscale
2. Start the daemon bound to the Tailscale IP:
TS_IP=$(tailscale ip -4)
vide daemon start --host $TS_IP --port 8080
The Tailscale IP (100.x.x.x) is in the CGNAT range -- never routed on the public internet. Your daemon is only reachable by devices on your tailnet.
3. Connect from your laptop:
vide connect 100.64.12.34:8080
# Or with MagicDNS:
vide connect my-vps.my-tailnet.ts.net:8080
4. (Optional) Install as a system service:
# Auto-start the daemon on boot
vide daemon install
vide daemon start # Start the daemon (foreground)
vide daemon start --detach # Start in background
vide daemon stop # Stop the daemon
vide daemon status # Check if running
vide daemon install # Install as system service (launchd/systemd)
vide daemon uninstall # Remove system service
Laptop (TUI) VPS
│ │
│── vide connect ──────────────► │ vide daemon (port 8080)
│ │ ├── session A (vide_server :54321)
│◄─── WebSocket stream ─────────│ ├── session B (vide_server :54322)
│ │ └── session C (vide_server :54323)
The daemon spawns a separate vide_server process per session for full isolation. Sessions persist across daemon restarts and survive client disconnects. Multiple TUI clients can connect to the same session simultaneously.
Vide doesn't call models directly -- it orchestrates full agent frameworks via their SDKs. Currently built on Claude Code. Codex CLI and Gemini CLI are on the roadmap. Same agent teams, same workflows, always using the best each framework has to offer.
.md files into .claude/agents/ to define your own specialists