Uploaded February 2026 | Updated September 2026, 3 weeks ago
Running one Claude Code agent is easy — managing five at once across multiple businesses and repos is chaos. Here's the custom Avalonia desktop app I built to track, organize, and control every AI agent from one screen.
After watching, you'll understand how to structure a multi-agent AI workflow, manage parallel coding tasks without losing track, and see what's possible when you build your own tooling around Claude Code.
In this video:
- The problem with managing multiple Claude Code agents in terminals and the desktop app
- Building a custom Avalonia UI to launch, name, color-code, and track agent sessions
- Real-time status indicators: idle, busy, locked — with sound notifications
- Grouping agents by repository and business so you never lose context
- Integrated to-do lists, bug tracking, and scratch notes per session and per repo
- Breaking out integrated terminals to independent windows without losing state
- One-click Git commit with AI-analyzed changelogs across multiple repositories
- Syncing Claude Code settings, skills, and CLAUDE.md between machines
When you start running five or more Claude Code agents simultaneously, the default terminal and desktop app fall apart — no status visibility, no titles, no way to know which agent is stuck waiting for permissions and which just finished. This video walks through every pain point of managing mass AI agents and demonstrates a purpose-built Avalonia application that solves each one. Color-coded sessions let you spot agent state at a glance across multiple monitors. Integrated notes and to-do lists mean you never forget what you told an agent to do. And the built-in Git workflow automates commits, pushes, and pulls across every repo. Whether you build your own management tool or adapt these ideas, the lesson is clear: treat AI agents like a team of developers and build the project management layer to match.
00:00 Introduction — The multi-agent problem
00:54 Claude Desktop app limitations
05:38 Terminal limitations — even worse
09:10 The custom Avalonia management app
10:02 Settings and quick-launch from Git repos
11:00 Recents, pinned sessions, and one-click sync
12:25 Spinning up multiple named, color-coded agents
14:00 Managing agents across different businesses
15:00 Detecting locked/stuck agents instantly
16:02 Titles, latest messages, and session history
17:00 Breaking out terminals to independent windows
19:00 Organizing and scheduling work efficiently
19:35 Integrated to-do lists per repository
22:00 Running to-dos as new agent tasks
24:00 To-dos shared across terminals in the same repo
25:30 Adding a bugs tab — live feature development
27:00 Reviewing AI work — page animations
30:00 Iterating on animation quality with AI
33:00 Bug tracking and persistence between sessions
37:00 Scratch notes, folder notes, and session notes
39:00 Automated Git commits with AI code analysis
41:00 Settings sync, skills management, and deployment
42:50 Closing thoughts — AI is a team, manage it like one
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Support me in my journey to giving back to the industry all my knowledge and helping the world with what I do. Spreading knowledge to those who cannot afford an education and helping those who want to better themselves.
Avalonia UI
avaloniaui.net
Source Code to Series
github.com/angelsix/youtube/tree/develop/Avalonia%20BatchProcess
Support what I do and pay for me to survive
angelsix.com/donate
Live Chat at Discord
discordapp.com/invite/eHr5BMk
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Running one Claude Code agent is easy — managing five at once across multiple businesses and repos is chaos. Here's the custom Avalonia desktop app I built to track, organize, and control every AI agent from one screen.
After watching, you'll understand how to structure a multi-agent AI workflow, manage parallel coding tasks without losing track, and see what's possible when you build your own tooling around Claude Code.
In this video:
- The problem with managing multiple Claude Code agents in terminals and the desktop app
- Building a custom Avalonia UI to launch, name, color-code, and track agent sessions
- Real-time status indicators: idle, busy, locked — with sound notifications
- Grouping agents by repository and business so you never lose context
- Integrated to-do lists, bug tracking, and scratch notes per session and per repo
- Breaking out integrated terminals to independent windows without losing state
- One-click Git commit with AI-analyzed changelogs across multiple repositories
- Syncing Claude Code settings, skills, and CLAUDE.md between machines
When you start running five or more Claude Code agents simultaneously, the default terminal and desktop app fall apart — no status visibility, no titles, no way to know which agent is stuck waiting for permissions and which just finished. This video walks through every pain point of managing mass AI agents and demonstrates a purpose-built Avalonia application that solves each one. Color-coded sessions let you spot agent state at a glance across multiple monitors. Integrated notes and to-do lists mean you never forget what you told an agent to do. And the built-in Git workflow automates commits, pushes, and pulls across every repo. Whether you build your own management tool or adapt these ideas, the lesson is clear: treat AI agents like a team of developers and build the project management layer to match.
00:00 Introduction — The multi-agent problem
00:54 Claude Desktop app limitations
05:38 Terminal limitations — even worse
09:10 The custom Avalonia management app
10:02 Settings and quick-launch from Git repos
11:00 Recents, pinned sessions, and one-click sync
12:25 Spinning up multiple named, color-coded agents
14:00 Managing agents across different businesses
15:00 Detecting locked/stuck agents instantly
16:02 Titles, latest messages, and session history
17:00 Breaking out terminals to independent windows
19:00 Organizing and scheduling work efficiently
19:35 Integrated to-do lists per repository
22:00 Running to-dos as new agent tasks
24:00 To-dos shared across terminals in the same repo
25:30 Adding a bugs tab — live feature development
27:00 Reviewing AI work — page animations
30:00 Iterating on animation quality with AI
33:00 Bug tracking and persistence between sessions
37:00 Scratch notes, folder notes, and session notes
39:00 Automated Git commits with AI code analysis
41:00 Settings sync, skills management, and deployment
42:50 Closing thoughts — AI is a team, manage it like one
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Support me in my journey to giving back to the industry all my knowledge and helping the world with what I do. Spreading knowledge to those who cannot afford an education and helping those who want to better themselves.
Avalonia UI
avaloniaui.net
Source Code to Series
github.com/angelsix/youtube/tree/develop/Avalonia%20BatchProcess
Support what I do and pay for me to survive
angelsix.com/donate
Live Chat at Discord
discordapp.com/invite/eHr5BMk
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
