DevPulse

Developer community research tool. Searches Reddit, Hacker News, and GitHub Discussions for recent developer conversations on any topic.

Why This Exists

Marketing content that references "what developers think" usually comes from guessing. DevPulse replaces guessing with data. It pulls real conversations from where developers actually talk (Reddit, HN, GitHub) and surfaces sentiment, pain points, and trends.

I built this as part of JakeOS, my AI-augmented marketing operating system, to fuel content research, competitive intelligence, and Reddit seeding monitoring at LayerLens.

Quick Start

pip install -r requirements.txt

# Basic search
python3 devpulse.py "AI evaluation tools"

# Deep scan with 60-day window
python3 devpulse.py "LLM benchmarking" --depth deep --days 60

# Reddit-only, last 2 weeks
python3 devpulse.py "Claude vs GPT" --sources reddit --days 14

# JSON output for programmatic use
python3 devpulse.py "vector databases" --output json

CLI Options

python3 devpulse.py "topic" [options]

Options:
  --depth     Search depth: quick, default, deep (default: default)
  --days      Time window in days (default: 30)
  --sources   Comma-separated: reddit, hackernews, github (default: all)
  --output    Output format: text, json (default: text)

Data Sources

Source	API	Auth Required	What It Searches
Reddit	ScrapeCreators API (paid) + free JSON fallback	Optional	Posts + top comments across auto-discovered subreddits
Hacker News	Algolia API (free)	No	Stories filtered by date, points threshold, with comment threads
GitHub	REST Search API	Optional (higher rate limits)	Issues, PRs, and repos by stars and creation date

Without API keys, Reddit falls back to public .json endpoints and GitHub allows 10 requests/minute unauthenticated. Everything works out of the box.

How It Works

Query Input
    |
    v
Query Normalization (extract core subject, expand synonyms)
    |
    v
Multi-Phase Search
  - Global search across all sources
  - Targeted subreddit/domain discovery from initial results
  - Concurrent requests across sources
    |
    v
Relevance Scoring
  - Token overlap with stopword filtering
  - Synonym expansion (ai -> artificial+intelligence, eval -> evaluation)
  - Phrase bonuses for exact matches
  - Low-signal word detection
    |
    v
Deduplication + Date Filtering + Engagement Sorting
    |
    v
Comment Enrichment (top posts get comment threads, depth > "quick")
    |
    v
Structured Output (text report or JSON)

Depth Levels

Depth	Search Breadth	Comment Enrichment	Best For
quick	Single-pass search	None	Fast pulse check
default	Multi-pass with subreddit discovery	Top 3 comments per post	General research
deep	Exhaustive multi-pass	Full comment threads	Deep competitive intel

Configuration

Create ~/.config/devpulse/.env (optional):

# Enhanced Reddit access (optional, free fallback available)
SCRAPECREATORS_API_KEY=your-key-here

# Higher GitHub rate limits (optional, 10 req/min without)
GITHUB_TOKEN=your-token-here

Use Cases

• Content research: Find what developers actually care about before writing
• Competitive intelligence: Monitor mentions of competitors across communities
• Trend spotting: Identify emerging topics before they hit mainstream
• Reddit seeding: Track how your posts and mentions spread
• Product research: Discover pain points and feature requests in your domain

Part of JakeOS

DevPulse is one component of JakeOS, an AI-augmented operating system for solo marketing. Other components include AI Marketing Skills and Buffer MCP.

License

MIT