Stable L1 + Shyft Identity + RMT Reputation — The agent trust layer that nobody else has assembled
reputation_score field with no mechanism to populate it. Our PageRankOracle fills that field.Stable built the fastest agent L1. Shyft built identity verification. We built reputation scoring. Each works alone. Together they create something no one else has — programmable trust at the consensus level.
Stable provides the infrastructure for agents to transact. Shyft provides the identity layer to verify who those agents represent. RMT provides the reputation layer to determine which agents deserve trust. No single team built all three. But the interfaces align.
Key: Each layer addresses a distinct question. Stable asks "can this agent transact?" Shyft asks "who is this agent?" RMT asks "should this agent be trusted?" Only the combination answers all three.
Each Stable module has a natural integration point for our identity and reputation layer.
| Stable Module | Our Integration | Result |
|---|---|---|
reputation_score field |
PageRankOracle.getScore() |
Every agent scored 0-10000 |
| Agent Intent Lane | Reputation-sorted tx ordering | Trusted agents get priority |
| ConditionalAuthorization | "counterparty rep > threshold" | Autonomous trust gating |
| x/agent-pay | Reputation-tiered payment limits | Rep = payment collateral |
| x/agent-mcp | DomainRegistry + resolver | Rate tools in compliance scope |
| x/agent-proof | MachineConsentHelper + TEE | Device liveness proofs |
| ZK precompile | Groth16 at 100-150K gas | Privacy-preserving rep proofs |
The distinction: Tempo solves payments. We solve trust. They tell you an agent can pay. We tell you an agent should be trusted.
Compliance-gated agent reputation is the first use case where no competitor can follow. Banks, insurance firms, and financial institutions deploying AI agents need every box checked simultaneously.
Why Tempo cannot serve this market: Stripe KYC is user-level, not agent-level. There is no mechanism to assign compliance scope to individual agents, no reputation scoring within those scopes, and no way to gate transactions on earned trust rather than payment credentials alone.
Reputation and staking create a self-reinforcing loop. RMTToken serves as a citation bond — stake when citing, get slashed if fraudulent.
We fill the empty reputation_score field in Stable's agent accounts with validated PageRank — turning the fastest agent L1 into the only chain where AI agents have Sybil-resistant, compliance-scoped trust.
getScore(agentAddress) — one call returns 0-10000 reputation backed by PageRank. Build agent-to-agent trust into any contract without running your own reputation infrastructure.
Individual agent KYC, compliance cohorts, quality-of-work reputation per context, autonomous transaction gating on identity + compliance + earned reputation. Full audit trails.
Cold start on chain 988. Need ~50+ active agents for meaningful PageRank scores. First epochs will be sparse. Mitigant: ERC-8004 bridge enables 24K agent passive import to seed the graph.
Tempo's distribution advantage. $500M funding and Stripe's merchant network give them reach we cannot match on capital alone. Our advantage is depth (trust infrastructure), not breadth.
Oracle centralization. Single operator for the PageRank oracle at launch. Gelato Web3 Functions multisig path is designed but not yet deployed. Centralization is a credible attack surface until decentralized.
Cross-chain score consistency. Merkle bridge for reputation portability is designed but not built. Scores are chain-local until the bridge ships.
Stable team integration speed. The full value of this stack depends on Stable reading reputation scores at the consensus layer. We control the contracts; we do not control their integration timeline.