token-gated API access

hold an NFT, access the cheapest LLM API without per-request payment.

the problem with per-request payment

x402 is brilliant for cold-start agents: no account, no API key, just pay. But for repeat customers making hundreds or thousands of calls per day, it is wasteful — every request triggers a fresh signed transfer and an on-chain settlement, even though the agent's intent is the same every time: "I want to use this service."

What if holding a token was the proof? Not a password, not an API key — an on-chain, verifiable, self-custodied signal that says: I am a committed member of this network.

how the prototype works

A single balanceOf call against a configurable ERC-20 or ERC-721 contract on Base determines eligibility. The check is:

• Read-only — no gas, no signing, no state change. Just an eth_call.
• Cached — a wallet's eligibility is cached for 60 seconds (configurable). Repeat callers incur near-zero RPC overhead.
• Fail-closed — if the RPC is unreachable or returns garbage, the wallet is denied, never granted access for free.
• Composable — works with any ERC-20 or ERC-721: a governance token, a membership NFT, a loyalty pass, a community badge.

When eligibility is configured, callers include their wallet via the X-Wallet header. Holders skip x402 entirely; non-holders fall through to the normal payment flow.

what a holder gets

the closed loop: NFT + cache flywheel

This is where it gets interesting. The NFT holder's wallet is simultaneously:

• Their payment identity (no API key, no account, just a wallet)
• Their SRP reward account (cache writes and hits mint SRP directly to their wallet)
• Their eligibility proof (holding = access, no per-request fee)
• A Sybil-resistant signal (to farm SRP you must hold the token, which costs capital)

And if the NFT sale revenue flows into the SRP rebate pool, every new member directly funds the flywheel for everyone:

agent buys NFT → rebate pool grows → SRP value rises
agent uses gateway → creates/uses cache → earns SRP
agent's SRP appreciates → effective cost drops → uses gateway more
more usage → more cache hits → more savings → more SRP value

what we need from the community

This is a prototype. We have not deployed a contract yet. Before we do, we need answers to real questions. That's what the feedback board below is for.

open questions

1. Should the token be an ERC-721 membership NFT (one per wallet, fixed price) or an ERC-20 governance token (variable balance, tradable)?
2. Should holders pay nothing at all (pure gate), or should they get a discount on per-request pricing (gate + revenue)?
3. Should there be multiple tiers (e.g. bronze/silver/gold NFTs with different rate limits and discounts)?
4. Should NFT sale revenue go to the SRP rebate pool, to operational costs, or a split?
5. Should there be a stake-and-slash mechanism (holding at risk, slashed for abuse) or is simple holding sufficient?
6. What price point makes the NFT accessible to independent developers but expensive enough to deter Sybil farms?
7. Should SRP become the on-chain token that grants eligibility, closing the loop completely?

community feedback

4 pieces of feedback so far. Your identity is a salted hash — raw handles/IPs are never stored.

recent feedback

related: vote on SRP / RevNet direction · how the cache engine works · live system status · API docs · about surp