Launched this week
LunoPeak
Your AI dev environment, at a glance
2 followers
Your AI dev environment, at a glance
2 followers
LunoPeak is a local-first desktop dashboard for developers using Claude Code and Codex. Watch sessions stream live, replay any past session step-by-step, and track per-model, per-repo spend with budget alerts and rate-limit windows in your tray. It also surfaces every repo, hook, skill, agent, and CLAUDE.md on your machine, plus hygiene fixes for stale branches, leftoverenv files, and ports. No account. No telemetry. API keys in the OS keychain.
Hey Product Hunt! 👋
I built LunoPeak because my AI coding workflow had outgrown the terminal. I had Claude Code running in one repo, Codex in another, three different CLAUDE.md files I couldn't keep track of, no idea what I'd spent this week, and a 5-hour usage window I kept blowing through mid-task.
LunoPeak is the cockpit I wanted: one tray icon, 27 live views, everything local.
A few things that make it different:
• Multi-agent, side by side. Claude Code and Codex sessions both stream into the Live view, bucketed per-process so they never cross-contaminate even when they're running in the same repo.
• Replay any session. Every prompt, tool call, and result — with cost and tokens on every step.
• Budgets that actually warn you. Per-model, per-repo, per-day spend, with alerts before you blow past your limit.
• Rate-limit windows in your menu bar. 5-hour and weekly Claude/Codex usage with a countdown to reset, so you can pace work with the app.
• Local-first, end of story. No account, no telemetry. API keys live in the OS keychain (Keychain / Credential Manager / Secret Service). The only outbound traffic is provider APIs when you use the Assistant, and the update check if you opt in.
• Hygiene that means hygiene. Stale branches, leftover stashes, missing CLAUDE.md files, ungitignored .env, ports stuck listening — surfaced with one-click fixes (or a one-liner you can run yourself).
It's free and works on macOS 10.15+, Windows 10+, and Linux (deb / rpm / AppImage). Built on Tauri 2 + React 19 — installs are ~150 MB and idle at ~120 MB of RAM.
Would love to hear what you'd want to see in the dashboard next. Reply with the metric, view, or workflow that's missing from your AI coding setup and I'll dig in.
🚀 LunoPeak 1.4 is live — a lot has changed since launch day.
When we shipped on Product Hunt at 1.2, LunoPeak was a local-first dashboard for Claude Code and Codex. Three releases later, it's grown up:
• 🧠 Ollama is now a first-class assistant provider. Run the chat on local models — no API key, no account, no cloud. Tool-calling works end-to-end on Llama 3.2, Qwen 2.5 Coder, Mistral. If your Ollama daemon is running, it's auto-adopted as the default before you even open Settings.
• 🎯 Cursor joined the lineup. Claude Code + Codex + Cursor — all three show up in Live, Setup, Repos, and Settings. One window, every agent.
• 🔀 Swap models from the header. Combined provider+model picker — Claude Sonnet → local Qwen → GPT and back, without leaving the chat.
• 🧩 Drag-and-drop dashboard. Rearrange your cards, layout persists.
• 📊 GitHub PRs in Work Graph, per-file dirty state on Repos, and a month-to-date spend that actually means MTD.
• 🔄 Dashboards refresh themselves now — 60-second cadence, no more frozen numbers.
Still local-first. Still no account. Still no telemetry. Still free.
👉 Grab 1.4.0: https://github.com/idevtim/lunop...