SignalTube

SignalTube turns long-form content into role-aware articles and presentations that help you understand what matters faster.

🎥 Demo

Experience how SignalTube adapts content to your role.

Watch the SignalTube demo

Download

The easiest way to try SignalTube is to download the macOS app from the latest GitHub release:

Download SignalTube for macOS

Current build:

• macOS Apple Silicon
• local-first desktop app
• works with local Claude, Codex, or Gemini CLI sessions

Note: the current DMG is an early public build. If macOS shows a security warning, open it using right-click > Open. A fully signed and notarized build is planned.

Why I built it

I have a constantly growing Watch Later list on YouTube. A lot of it is podcasts, interviews, AI announcements, technical videos, and research discussions. I also save articles from Medium and Substack.

The problem was not only that there was too much to watch or read. The bigger problem was that even when I tried to consume it, the useful signal was often hard to understand, especially when the topic was technical or outside my direct field.

Generic AI summaries helped a little, but not enough. If the original topic is complex, a generic summary can still leave you wondering: "What does this actually mean for me?"

SignalTube is built around that exact gap.

What SignalTube does

Paste a YouTube, Medium, or Substack link, choose a role or perspective, and choose the output you want.

SignalTube then turns the source into:

• Article · Short Dive for quick understanding
• Article · Deep Dive for a richer long-form read
• Presentation for a slide-style version you can scan or share

The core idea is simple: SignalTube does not just summarize content. It interprets long-form content based on who is reading it.

What makes it different

Most summarizers compress the same source in the same way for everyone.

SignalTube is built for role-aware understanding:

• a design manager can extract what an AI research discussion means for team direction
• a founder can use the same source to understand where the market or industry is heading
• a UX designer can focus on product implications, user behavior, and interface tradeoffs
• a developer can look for implementation signals, architecture ideas, and tooling direction
• a manager can turn dense material into something easier to share with a team

That makes the output more usable, not just shorter.

Supported inputs

• YouTube
• Medium
• Substack

Supported outputs

• Article · Short Dive
  A concise role-aware memo with summary, key ideas, key concepts, and relevance.

• Article · Deep Dive
  A richer editorial version for when you want nuance and deeper context.

• Presentation
  A slide-style quick-read output for reviewing or sharing.

Local AI providers

SignalTube works with AI tools already installed on your machine:

• Claude CLI
• Codex CLI
• Gemini CLI

The app invokes these tools in the background through local CLI calls. It does not create a visible saved chat thread in those tools.

Role-aware output

SignalTube supports predefined roles such as:

• HAI Designer
• UX Designer
• Developer
• I'm a kid

It also supports custom roles. You can define your own role by describing:

• what you do day to day
• what decisions or outcomes matter most
• what perspective SignalTube should use when interpreting content
• any extra context that would make the output more useful

Screenshots

Homepage view

Article view

Key Ideas

Presentation

Library

Open source notes

This repository includes simple generic prompts so the app remains usable and understandable as an open-source project.

Private distribution builds may use more specialized prompt variants. The public prompts are intentionally simpler, but they still demonstrate the product flow and produce usable outputs.

Setup from source

Requirements

• macOS
• Node.js 20 or newer
• npm
• one supported local provider CLI installed and signed in:
  • Claude CLI
  • Codex CLI
  • Gemini CLI

Install

git clone https://github.com/vvmahesh0/SignalTube.git
cd SignalTube
npm install

Run the desktop app

npm run build
npm run desktop

Run web development mode

npm run dev

Local-first behavior

• generated memos are saved locally on your machine
• repo clones do not include anyone else's library or history
• new users start with a fresh empty library
• no server-side API key is required by the app itself

Project structure

• app/ - routes and API endpoints
• components/ - UI screens and shared components
• lib/ - ingestion, storage, providers, prompt loading, and parsing
• Article Short Dive Prompt.md
• Article Deep Dive Prompt.md
• Presentation Prompt.md

License

MIT