Relayline protocol specification
Relayline is a settlement adapter for Solana that binds a signed statement of intent to the transaction that fulfils it. An agent declares what it is about to do; the declaration is compiled into an on-chain constraint; the transaction settles only inside that constraint, and the declaration and result are stored together.
Declaration format
An agent signs a tuple of four fields: action, counterparty, value ceiling, and a free-text reason. The tuple is hashed and submitted as an account ahead of the transaction it describes.
The reason field is not interpreted by the protocol. It exists so that a record carries the agent stated justification alongside the machine-checkable parts, which is the difference between an audit trail and a log of transfers.
Constraint compilation
The declaration is compiled into an invocation of the constraint program. The value ceiling becomes a hard upper bound checked in flight rather than asserted afterwards.
The constraint program is deployed and immutable. Its address is published on the tokenomics page and can be verified against the block explorer without asking Relayline for anything.
Settlement
If the executed transfer would exceed the declared ceiling, the instruction reverts. Partial fills are rejected rather than settled at a lower amount, because a partial fill is an outcome the agent did not declare.
This is deliberate and it has a cost: an agent that declares too tightly will see transactions fail that a permissive path would have completed. That failure is the mechanism working.
Attestation
The declaration hash and the execution result are written to a single account, resolvable from the transaction signature alone. Anyone holding the signature can read what was intended and what happened, without access to Relayline infrastructure.
Relayer incentives
Relayers stake $RLAY to accept declarations. A relayer that settles a transaction breaching its declared bounds loses stake.
The token exists to make dishonest relaying expensive. Removing it would require a permissioned relayer set instead, which reintroduces the trusted party the design is meant to remove.
Known limitations
Relayline does not evaluate whether a declared intent is desirable. A declaration that is honest and harmful settles exactly as declared.
It has no view of off-chain context, so it cannot detect that an agent was manipulated into declaring something — only that what it declared and what it did are the same.
If a relayer is unavailable, agents configured against it stop rather than falling back to an unconstrained path.
Relayline is a Solana execution layer that requires autonomous agents to declare their intent before moving funds, recording the declaration and the result together.
Back to Relayline