Identify the contribution
Write down what the project is measuring: verified compute, useful data, model evaluation, API calls, agent uptime, feedback or simple referrals. Prefer programs where the action connects to a real product need.
Protect data and credentials
Do not upload confidential prompts, proprietary data, customer information or API keys for a speculative reward. Review licenses and retention terms before contributing datasets or outputs.
Check compute economics
For GPU or node programs, estimate electricity, hardware wear, bandwidth, collateral, slashing, uptime and payout uncertainty. A point multiplier can hide a negative operating margin.
Verify agent permissions
An agent that can sign transactions, access an exchange or call APIs should use scoped keys and strict spending limits. Do not grant broad wallet authority to an unaudited bot.
Separate product metrics from token rumors
Track shipped features, documentation and verifiable technical work. Avoid treating investor names, social engagement or the word “AI” as a substitute for a functioning system.