Project Launchpad is an elite, offline-first multi-agent AI pipeline designed to automate developer-marketing workflows. By bridging the gap between engineering outputs and content writing, Launchpad parses local project code, screenshots, and developer logs, crawls real-time web references using a built-in Puppeteer search crawler, and synthesizes high-engagement Substack articles and plain-text social content tailored to your authentic writing style.
Built inside a dark, premium macOS-style desktop application following Apple's Human Interface Guidelines (HIG).
Developer advocacy and marketing often feel like chores that pull you away from coding. Launchpad acts as your personalized, automated agency:
- Structured Marketing Planner: Formulates ideas using classic narrative structures (Deep Tech Dives, Builder's Struggle, Quick Hack & Tutorial).
- Free Local Crawler: Spawns Chromium headlessly to scrape web pages and synthesize search reports without relying on expensive SaaS search credits.
- Persona and Learnings Engines: Automatically parses and adapts to your tone, emojis threshold, avoid-lists, and historical analytics feedback.
Launchpad chain-links separate focused AI agents (Substack Writer, Humanizer, Image Positioner, LinkedIn Writer) to review, revise, and place media contextually. It maintains absolute style alignment by referencing a centralized persona.json and a running ledger of rule-of-thumb constraints (learnings.txt).
Drafts are organized dynamically inside projects/<project-name>/output_files/<model-name>/ subfolders. This allows you to A/B test outputs generated by Google Gemini and local Ollama side-by-side without file overwrite conflicts.
Features frosted glass dropdown selection triggers, custom spring kinetic menus, collapsible drawer layouts, and real-time pulsing neuron nodes representing agent execution states.
launchpad-app/
├── server.js # Express API server (Puppeteer crawlers & LLM handlers)
├── client.js # Client state, event listeners, & tab handlers
├── index.html # App container & macOS desk layout panels
├── style.css # Custom Apple HIG stylesheet & neural keyframes
├── CHANGELOG.md # Highly polished release version notes
├── projects/
│ ├── persona.json # Centralized writing style configurations
│ ├── learnings.txt # Continuous system rules of thumb
│ ├── linkedin_good_posts.json # Analyzed style reference library
│ └── substack_good_posts.json # Crawled substack reference structures
└── node-env/ # Headless Chromium & Node execution binaries
- Node.js (v18+)
- An active Google Gemini API key or local Ollama server running Qwen2.5-Coder.
- Clone the repository:
git clone https://github.com/Mudit01100001/project-launchpad.git cd project-launchpad - Install dependencies:
npm install
- Configure your environment parameters:
cp .env.template .env # Edit .env and supply your API keys
Start the server in your workspace context:
node server.jsOpen http://localhost:3000 in your browser.
Refer to the CHANGELOG.md to inspect detailed release notes and system changes.