MultiTable – I built a dashboard so I could vibe-code from my phone
One tab for Claude Code, Codex, and Aider when tmux wasn't built for agents.

Dishonored shameboard for fake lifters sets this apart from Hevy and Strong.
Gym goers, fitness enthusiasts
Hevy · Strong · FitNotes
- "this exercise hurts my shoulder at the end of the movement, remind me proper form and cues"
- "how should I modify my workout today based off last week's reps and weights"
- `/generate` a bouldering-focused workout routine
The app is free, your data is fully exportable (no walled-gardens), and there's an anonymous global leaderboard for motivation, including a "Dishonored" shameboard for anyone caught faking their numbers.
I decided to 100% vibe-code this project, a part of it is halluci-coded. I purposely haven't touched a single line of code, though I've reviewed most of it. The project started as just my own workout split, but friends wanted to try it so I added routines, then feature requests started coming in and I basically just forwarded them to my CLI agents.
Vibe-coding is far from perfect, but since I'm a daily user of my own app I catch bugs immediately, sometimes literally in the gym while SSHing from my phone between sets.
---
Tech stack: - FastAPI + SQLite on a Linux box - React PWA (works offline) - AI coach powered by Claude via Anthropic API - Telegram OAuth + Resend magic-link auth - Sentry for error tracking, Foursquare Places for gym location
The whole thing was built with Claude Code. Vibe-coded with Claude, and the AI coach is also Claude. It's Claude all the way down.
---
What do you hate about your current fitness tracker? I might fix that.
One tab for Claude Code, Codex, and Aider when tmux wasn't built for agents.
Closes the loop: agents read eval traces to fix their own regressions.
Yet another AI code reviewer when Cursor and Continue already dominate this space.
Clone of Hitster built entirely via voice prompts to Claude from a car.
Agent-agnostic remote control for Claude Code, Cursor, and Copilot CLI sessions.
One-command hardening script for AI agents when official docs are scattered.