23blocks-OS

AI Maestro: A Development History — Data & AI agent for Claude Code

Data & AI community intermediate

From "AI Agents Web Dashboard" to multi-host agent orchestration in 110 days.

How to install AI Maestro: A Development History

This entry records only its repository, not the path inside it, so there is no exact command to give. Open 23blocks-OS/ai-maestro and copy the folder into ~/.claude/agents/.

What AI Maestro: A Development History does

*From "AI Agents Web Dashboard" to multi-host agent orchestration in 110 days.* ---

Alternatives in Data & AI

Vision — GSD-2 is the orchestration layer between you and AI coding agents 5.7k ★
Claude Sub Agent — AI-driven development workflow system built on Claude Code Sub-Agents 566 ★
General Fullstack Developer — Use this agent proactively when you need end-to-end feature development that spans multiple layers of the appl 248 ★

README

AI Maestro: A Development History

*From "AI Agents Web Dashboard" to multi-host agent orchestration in 110 days.*

The Numbers

Metric	Value
First Commit	October 9, 2025
Current Version	v0.19.31
Total Commits	628
Pull Requests	137
Days of Development	110
Primary Developer	Juan Pelaez (740 commits)
AI Co-Author	Claude (10 commits)

Chapter 1: The Beginning (October 9-14, 2025)

**The First Day - October 9, 2025**

It started with a simple idea: manage multiple Claude Code agents from one dashboard.

d317980 - Initial commit: AI Agents Web Dashboard
765be95 - Add collapsible notes, redesign sidebar with dynamic colors
1377ee3 - Rebrand to AI Maestro and prepare for open source launch

Within hours, the project had a name ("AI Maestro"), a logo (the constellation), and was deployed to GitHub Pages. The CNAME was created, deleted, and recreated within minutes—the first of many rapid iterations.

**The Terminal Wars - October 10-12**

The first real challenge emerged: getting terminal rendering right.

v0.1.2 - Terminal scrollback improvements
v0.1.3 - Add scrollback workarounds and terminal improvements
v0.1.4 - Scrollback improvements and workarounds
v0.1.5 - Fix regex pattern error and restore session content

Four versions in two days, all fighting with xterm.js scrollback, line rendering, and text selection. The commits reveal the struggle:

"Fix Claude Code terminal scrollback and line rendering issues"
"Fix terminal scrollback buffer and improve performance"
"Fix terminal rendering issues"

**PR #1: Messaging - October 12**

Just three days in, the first major feature landed: agent-to-agent messaging.

c702a7a - Add agent-to-agent messaging system v0.2.0

This would become one of AI Maestro's defining features—but not without more terminal battles:

v0.2.1 - Fix terminal scrollback, text selection, and line ending issues
v0.2.2 - Fix terminal rendering issues
v0.2.3 - Fix terminal scrollback buffer and improve performance

Chapter 2: Architecture Evolution (October 13-31, 2025)

**The Tab-Based Revolution - PR #2**

October 13 brought a fundamental architectural change:

4c87d7e - Refactor to tab-based terminal architecture

The old approach: mount/unmount terminals on agent switch. The new approach: all terminals mounted simultaneously, toggle visibility.

This eliminated race conditions and made agent switching instant. The CLAUDE.md documentation was updated to memorialize this decision forever.

**The First Crisis: PTY Leak - PR #4**

October 17, 2025. The first real crisis:

5106455 - Fix critical PTY leak bug causing system resource exhaustion (v0.4.0)

PTY handles weren't being cleaned up. The system would slowly exhaust resources until terminals stopped working. This bug would resurface months later (#104), proving that some battles are never truly won.

**Messages Get Serious - PRs #5-9**