tianyilt

tianyilt / claude-fleet

Public

Dashboard for orchestrating many concurrent Claude Code / Codex windows — triage, full-text search, skill/memory analytics

11
2
85% credibility
Found Jun 01, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Claude Fleet is a local web dashboard that helps people who use Claude Code with multiple windows open at once. It shows the status of every AI assistant window - whether it's actively working, waiting on you to approve something, stuck on a problem, or finished and ready to close. The tool reads data from your computer's local folders where Claude Code stores its conversations and session information; it never sends anything to the internet or changes any of your AI assistant's files. You can jump quickly to any terminal, search through all your past conversations, see which helper tools (called skills) each assistant used, and track how memory files were read or written. It's designed for developers who 'vibe code' with several AI helpers working simultaneously.

How It Works

1
🎨 You're doing vibe coding with many AI assistants

You have 5-7 Claude Code windows open at once, each working on different tasks, and you can't keep track of everything.

2
🚀 You open Claude Fleet in your browser

One click launches a beautiful dashboard showing all your AI assistants at once - what's working, what's stuck, what's waiting on you.

3
👀 You see the status of every window at a glance

Each card shows a colorful status: green for working, red for waiting on you, yellow for stuck, and gray for safe to close.

4
🔔 Permission alerts grab your attention

If any assistant is waiting for you to approve something, a persistent red bar appears so you never miss it.

5
You want to do something with an assistant
🎯
Jump to it

Click 'Focus' and your terminal tab instantly comes forward, exactly where that assistant is running.

📝
Review work

Ask a helper to check if everything looks good before closing.

🔄
Continue later

Resume or fork a session to keep working on it another time.

6
🔍 Search across everything you ever worked on

Type any word and find it across all your past conversations, even if it was just mentioned casually in a chat.

🎉 Your multi-window chaos becomes organized

With everything visible in one place, you confidently manage multiple AI assistants working together on your projects.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is claude-fleet?

Claude Fleet is a local web dashboard for developers who run multiple Claude Code or Codex sessions simultaneously. Instead of frantically switching between 5-7 terminal windows to figure out which one is stuck waiting for permission, which finished its task, and which is still grinding away, you get one unified view. The Python backend reads your existing session data from your home directory and serves a lightweight single-page frontend via FastAPI. It classifies each window's state (working, waiting, stalled, completed), shows real-time task hints, and lets you jump directly to any terminal tab with one click.

Why is it gaining traction?

The killer feature is the triage engine. It reads transcript data to understand what's actually happening -- not just whether a process is alive, but whether it's blocked on a permission prompt, has background tasks running, or has been idle for over an hour. The search is fast because it uses ripgrep across your transcript files, returning results in about 50 milliseconds with context snippets. There's also surprisingly deep analytics: you can see exactly how much your custom skills and memory files are being used across sessions, not just formal invocations but file reads, writes, and bash references too.

Who should use this?

If you're doing serious agent-heavy development with multiple concurrent Claude Code or Codex instances -- especially if you're building agents that use skills and memory extensively -- this dashboard solves a real coordination problem. Power users managing 3+ agent windows simultaneously will get the most value. Solo developers doing vibe coding with 1-2 agents probably don't need it yet.

Verdict

It's a clever, well-scoped tool that fills a genuine gap in the Claude Code workflow. The codebase has clean architecture and the README is excellent. At 11 stars it's clearly early-stage, and there are no visible tests -- so treat it as promising rather than production-ready. The 0.85% credibility score reflects a young project with limited community validation. Worth installing if you're running a multi-agent setup; otherwise, watch it from the sidelines.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.