Never a screenshot, unless it has to be. Product Hunt's own record decides, first.
A trust score is only worth anything if what it's built on can actually be checked. This is exactly how LaunchClan verifies that a pact was kept.
The method
Product Hunt's API, not a human squinting at a screenshot
The product's core claim is that verification doesn't rely on screenshots. Here's what that means in practice.
One query, unforgeable
With your own Product Hunt connection, we ask Product Hunt directly whether you voted for the specific post an obligation is for — using your token, so the answer reflects your account and no one else's.
A miss isn't assumed
If the direct check can't be run — you haven't connected Product Hunt, or Product Hunt's rate limit is hit — that's inconclusive, not a fail. We never penalise a trust score on an inconclusive result.
Screenshots are checked by AI first, a human second
Without a Product Hunt connection, you can submit a screenshot instead. It's reviewed automatically against the launch it's meant to prove — a confident read resolves it immediately, in either direction. Anything the AI isn't confident about goes to a human at /admin/review rather than being guessed at.
Every check leaves a record
Pass or fail, every real verification attempt is logged with its method and outcome, so a disputed result can be re-read against actual evidence rather than taken on trust.
Fairness
A trust score has to survive a dispute
Every rule below exists because penalising the wrong side, even once, makes the score worthless.
No link, no penalty
If the founder you backed never added their Product Hunt link, there was nothing you could have done. That obligation is waived, not counted as a no-show — the rule is fixed, not a judgement call.
Windows, not a single deadline
Each obligation opens on the partner's launch day and stays open for a fixed window. A missed commitment is only recorded once that window has fully closed.
One outcome per obligation
Once an obligation is verified, waived, or recorded as missed, re-checking it does nothing — settled outcomes can't be flipped by checking again, in either direction.
Rate limits are real
Product Hunt's API is complexity-priced, so during a busy launch window a verification check can be rate-limited. That returns as inconclusive — try again shortly, and it never costs you a trust penalty in the meantime.
Money never touches a vote
Verification works exactly the same way whether a launch is free or activated — there is no faster or weaker check to pay for. Support on LaunchClan is always a founder choosing to back another founder, and every commitment is verified against Product Hunt either way. Nothing here buys an upvote, a pact, or a trust score.
See it in the product
Connect Product Hunt and your next obligation is checked in one round trip, automatically.