Looking for design partners
Allow, challenge, or deny costly actions.
Without guessing which real customers to block.
Free-credit farming is the clearest case. Chitmark shows what abuse is costing you, then puts an allow, challenge, or deny decision in front of costly actions, without blindly blocking legitimate customers.
For teams where free grants, trials, or agent actions have real variable cost and enough volume to observe abuse within weeks.
PrincipleFail to challenge. Never to allow.
Works alongside Clerk, Auth0, Cloudflare, Stripe, and your existing auth stack.
Built for modern stacks
- TypeScript
- Python
- Edge
- CLI
- Next.js
- Express
The beachhead
Free-credit abuse has a dollar figure. Start there.
Run a read-only baseline on your own signup and usage data. Get farming cost, what slipped past existing controls, and whether enforcement is worth it. Nothing is blocked during the analysis.
Measure your actual burn
A read-only pass over your signup and usage data. Farming cost in dollars, what slipped past existing controls, and whether enforcement is worth it. The economics calculator estimates exposure; the baseline measures it.
Generic fraud tools say traffic looks suspicious. Chitmark returns a measured dollar figure from your own data.
Challenge uncertain actions
Legitimate customers keep moving. Actions that look like farming get a challenge before they consume the credits, trial, or quota you pay for. Uncertainty never becomes a free pass.
A weekly report of cost and outcomes
The weekly ledger shows what was blocked, what slipped, and what it cost. Review one documented change at a time.
Where the decision sits
Start at signup. Protect every costly action.
Signup is a common first surface, not the category. Put the same allow, challenge, or deny call wherever a grant has real variable cost.
Signup
Decide before a new account can draw free credits. A frequent first wire, still just one surface.
API key
Same verdict when a key unlocks quota or compute. Farming pays in friction; real builders keep moving.
Trial
Gate trial starts the same way. Harvest patterns show up as burn without conversion.
Expensive action
Extend to any consequential inbound action once the farming economics are clear. One loop, more surfaces.
How it works
Three verbs behind the decision.
Once the farming cost is real, the architecture is simple: verify at the grant, report the outcome, challenge when the call is unclear. When Chitmark cannot be sure, the action gets a challenge: a slow dependency never means a free pass. Runs alongside your existing controls.
Return an allow, challenge, or deny decision in under 50 milliseconds. The same inputs produce the same decision.
POST /v1/verifyConnect outcomes: credit burn, conversion, chargeback, back to the original action. Weekly review of what happened, one documented change at a time. Labels, not a billing stack.
POST /v1/feedbackAdd friction when a decision is not clear, using your existing method first (like verified email or delayed credit release). Cheap for one real user, expensive at farming scale.
POST /v1/challengeFree baseline, read-only
Get your free baseline.
Send us a CSV of signups and usage. Within a week, we'll show you:
- measured free-credit burn from farming
- the cohorts driving it
- what existing controls missed
- conversion contrast against clean traffic
Nothing is blocked during the analysis.
If the number is material, we'll discuss a pilot. If it isn't, we're done.
Prefer email? beta@chitmark.com