Your agent history is worth money.
Upload Claude, ChatGPT, and coding-agent traces. We scrub, price, and license them to AI labs, and you get paid when they sell.
Works with Claude · ChatGPT · Cursor · Codex · Gemini
Learn moreHow it works
From export to payout. In three steps.
- 01
Upload from any agent.
A Claude or ChatGPT export, the CLI pointed at your history, or a ZIP — Codex, Cursor, and Gemini too.
- 02
Scrubbed, then priced.
Secrets and PII are stripped on your machine, then every trace is scored and priced.
- 03
Licensed to AI labs.
Approved batches sell to labs and benchmarks. You keep 80% of every sale.
$
Runs locally. Secrets and PII never leave your machine.
- Eval lab$8.40/trace
- RL workflow$22.00/trace
- Code benchmark$75.00/trace
List your traces in one command.
What contributors say
I had 600 Claude sessions rotting in an export. Listed them in one command and made my first $300 the same week.
The scrub is the part I didn't want to build myself: secrets and PII are gone before anything leaves my machine.
Finally a use for my Codex debugging trails. Pricing is transparent and I keep 80% of every sale.
Pointed it at my agent history, got an estimate in seconds, approved a batch. That's the whole thing.
Estimates aren't payouts, you're paid only after an approved batch sells. How we handle your data →