How to vibe codeSubSignal

Prerequisites

• Node.jsfree

  Runtime environment for the Next.js web application and background workers.

• GitHubfree

  Version control and CI/CD triggers for deployment.

• Anthropic API keypay-as-you-go

  Powers semantic intent classification and draft reply generation.

AI coding tools

• Claude Code

  Best-in-class terminal agent for bootstrapping full-stack Next.js applications and writing robust background scraping/queuing logic.

• Cursor

  Ideal for iterative UI work, fine-tuning dashboard components, and inspecting database query diffs.

Build guide

1. 01Scaffold Project & Database Schema

   Initialize the Next.js project with Tailwind CSS, configure Turso SQLite via Drizzle ORM, and set up better-auth for single-user or multi-user access.

   Create a new Next.js 16 project using TypeScript and Tailwind CSS v4. Set up Drizzle ORM configured for Turso SQLite. Define database schemas for users, monitored_keywords (id, userId, keyword, subreddits, active), posts (id, redditId, title, url, subreddit, author, content, score, intentScore, matchedKeywordId, createdAt), and alerts. Integrate better-auth for local credential authentication. Verify the schema compiles cleanly and migration scripts run without errors.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

2. 02Reddit Ingestion Engine

   Build a robust ingestion module that polls Reddit JSON feeds or API endpoints with exponential backoff and request rate limiting.

   Create a backend worker service using Node.js and TypeScript that periodically queries Reddit search JSON endpoints and target subreddit RSS/JSON feeds. Implement strict rate-limiting, request deduplication by redditId, and error handling with exponential backoff to prevent blocks. Save newly discovered posts into the Turso database, ensuring old duplicates are ignored.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

3. 03Semantic AI Intent Scorer

   Integrate the Vercel AI SDK and Anthropic API to analyze incoming Reddit posts against user-defined product descriptions and keywords.

   Implement an intent scoring service using the Vercel AI SDK and Anthropic API (Claude Sonnet). For every newly ingested Reddit post, send the post title, body, and the user's product description to the LLM to output a JSON object containing: isMatch (boolean), buyingIntentScore (1-10), urgency (low/medium/high), and a brief rationale. Store these results back in the posts table.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

4. 04Dashboard Lead Finder & Filtering UI

   Build the main SaaS dashboard views to display, filter, and sort high-intent Reddit leads by urgency, score, and subreddit.

   Build a responsive SaaS dashboard UI using Tailwind CSS and Lucide icons in Next.js. Create a 'Lead Finder' view that lists ingested Reddit posts matching the user's criteria. Include filter controls for minimum intent score, subreddit selection, and date range. Add card components showing post metadata, AI rationale badges, and direct links to the Reddit threads.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

5. 05AI Reply Generator & Notifications

   Add contextual draft reply generation for individual leads and email notification dispatches via Resend for urgent matches.

   Add an AI Reply Generator modal to the lead detail view that calls Claude to draft 3 authentic, non-spammy engagement response options tailored to the specific subreddit's culture. Implement an alert dispatcher using Resend that triggers instant email notifications when a post with an intent score above 8 is indexed. Verify end-to-end flow from ingestion to email delivery.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

Cost vs paying for SubSignal