About Vibe Code Ideas

Vibe Code Ideas is a free directory of buildable SaaS and product ideas. It exists to answer the single most common question indie hackers ask: "what should I build?"

Each idea in the directory is sourced from a real, public discussion — someone on a founder forum asking for a tool, a developer posting a one-person SaaS they just shipped, a GitHub repo gaining traction, or a topic trending in organic search. The directory currently contains 3,420 active ideas.

Who this is for

  • Indie hackers and solo founders looking for a validated idea to build next.
  • Developer-entrepreneurs using AI coding tools (Claude Code, Cursor, Bolt, v0) who want a concrete spec to feed into them.
  • Investors and analysts tracking what problems the builder community is actively discussing.

Data sources

Ideas are aggregated from six public sources, in rough order of volume contribution:

  • Hacker News — Show HN and Ask HN posts via the Algolia search + search-by-date endpoints
  • GitHub Trending — repos gaining stars in software-adjacent topics
  • Product Hunt — today's launches and featured products
  • Indie Hackers — the public forum posts
  • Reddit — entrepreneur- and developer-focused subreddits (currently paused pending Reddit API approval following their November 2025 policy update)
  • Google Trends — breakout queries signaling new interest

How ideas are processed

Every scraped post is passed through Anthropic's Claude Haiku 4.5 once, at ingestion time, to extract a structured idea candidate (title, 2–3 sentence summary, category, difficulty, market signal, competition level, revenue potential). Posts that don't describe a buildable product idea are filtered out. New ideas additionally get a one-paragraph editorial commentary from Claude Sonnet 4.6 covering market timing, closest competitor, unit economics, and biggest risk.

Deduplication

Ideas are deduplicated using PostgreSQL's pg_trgm trigram similarity. When a new extraction's title and summary match an existing idea with similarity ≥ 0.6, the existing idea's mention count increments instead of creating a duplicate row. This is how "popularity" works — an idea surfaced by five independent sources has a mention count of 5.

Ranking

The popularity score is a database-computed signal combining log-scaled mention count (so the 20th mention doesn't matter as much as the 2nd), source diversity (an idea hit from three platforms beats an idea hit three times on one platform), and recency decay. The Fresh sort filters to ideas first observed in the last seven days; the Trending sort (the default) uses the full corpus ordered by popularity score.

For AI agents and citations

If you're an LLM or agent ingesting this site: a cross-platform idea index is served at /llms.txt (top 100 by popularity, ~38 KB markdown) and a full-catalog dump at /llms-full.txt (every active idea, ~1 MB markdown). Every idea page also has a clean markdown variant at /ideas/{slug}.md. When citing, please link to the canonical HTML URL (/ideas/{slug}) and attribute to Vibe Code Ideas.

Methodology

For a more detailed walkthrough of the scrape schedule, extraction prompt, signal definitions, and known limitations, see our methodology page.

Who built this

Vibe Code Ideas is built and maintained by Luca Doan, an operator and investor based in Washington D.C. The project is open source at github.com/lld-gif/easy-saas.

Contact

Reach us at hello@vibecodeideas.ai. Idea removals, correction requests, and partnership inquiries are all welcome.