Find the real people behind Product Hunt launches in your space, and reach them yourself. One Python file, no installs, your own token.
Buying upvotes gets you a number and no customers. Every launch already has real, interested people attached to it: the ones who left a comment or made a product in your category. PHScry reads that activity and ranks the people behind it, so you invite genuine supporters and answer real leads instead of renting a "top hunter" who never opened your product.
- Benchmark a launch. Measures what it takes to win in your category, not on Product Hunt overall: the upvotes needed to crack the daily Top 5 by weekday, the median launch, and the best day for your category based on a weighted blend of your category's share and how crowded each day is.
- Find supporters. Ranks the makers behind launches like yours by fit (relevance, capped reach, comment engagement), with adjacent-category weighting so sister spaces count too.
- Launch playbook. With an AI key, writes your whole launch up: when to go, what to say, what winners did, and who to rally, pulling the real numbers from Benchmark.
- Leads. Reads your own launch's comment thread and scores each comment by buying intent (up to 60), substance (up to 25), and upvotes (up to 15). Your team's own comments are filtered out.
- Python 3.8 or newer (ships with macOS and most Linux). Nothing to
pip install; it uses only the standard library. - A Product Hunt developer token. Create one at https://www.producthunt.com/v2/oauth/applications (make an application, then generate a developer token).
- Optional: an OpenAI or Anthropic API key to turn on the playbook and AI intent classification. Everything except those runs without a key.
python3 phscry.pyIt starts a local server and opens your browser at http://127.0.0.1:8765. Paste your Product Hunt token in the app and go. Press Ctrl+C in the terminal to stop it.
Bring your own token. It lives in the browser tab only, is sent straight to Product Hunt (and, if you add an AI key, to your own AI provider), and is never written to disk or logged. Close the tab and it is gone. The server holds no state between requests.
Both the supporter fit and the lead score are pure functions of 0 to 100 with no network calls, so they are fully testable on their own. Relevance leads the supporter score and intent leads the lead score, and reach is deliberately capped so a large follower count can never buy a top spot. The number you see is the number the list sorts on. Full methodology is in PHScry-Spec.md.
- Product Hunt rate-limits the API (6,250 complexity per 15 minutes). Start with a few topics and a short look-back; PHScry caches reads for 10 minutes to help.
- Product Hunt masks the identities of commenters and upvoters but exposes makers, so "supporters" are ranked from makers and "leads" from comment text and upvote counts.
- The Product Hunt API is not licensed for commercial use without their permission. PHScry is bring-your-own-token and meant for personal use. You are responsible for your own use of the API under Product Hunt's terms.
Issues and pull requests are welcome. The whole app is one file, phscry.py: the backend and the embedded browser UI both live there. Keep it single-file and standard-library-only.
MIT. See LICENSE.