Intents
The intended model for expressing what you want rather than how to get it — and an honest account of how little of it exists.
None of this is built. 계획됨
There is no intent endpoint, no quoting engine, no solver interface and no settlement path. Nothing on this page can be called today.
It is documented because the shape informs how the built parts are designed, and because you deserve to know where the network is heading. It is not documented so you can build against it. If you write code today that assumes any of this, that code will not work — not "will need adjusting", will not work, because there is nothing behind it.
The idea
An intent is a statement of what you want, not how to get it.
The distinction: a transaction says "call this function with these arguments". An intent says "I want this outcome, within these constraints" and leaves the route to whoever can find the best one.
Transaction: "route through A, then B, accept this slippage"
Intent: "I want at least X of this outcome, before this deadline"Whoever works out the route is a solver. Solvers compete; the user gets the outcome they specified or nothing at all.
Why the model is attractive
- The user does not need to be an expert in routing. They state a goal and a bound.
Competition is on execution quality. Solvers win by being better, not by being first in a queue.
Failure is bounded by the constraint. If nobody can meet it, nothing happens — which is a much better failure mode than partially happening at a bad price.
The planned surface
| Endpoint | Status | Would do |
|---|---|---|
POST /v1/dex/intent | 계획됨 | Submit an intent. |
GET /v1/dex/quote | 계획됨 | Ask what a solver would offer. |
POST /v1/dex/solve | 계획됨 | Submit a solution as a solver. |
GET /v1/dex/pools | 계획됨 | Read available liquidity. |
What the SDKs do about it
They raise, with the reason:
await ln.dex.quote();
// → NotYetWiredError [NOT_YET_WIRED]: dex.quote() is planned: no quoting system exists.
// → https://docs.libertynet.ai/statusln.dex.quote()
# → NotYetWiredError: [NOT_YET_WIRED] dex.quote() is planned: no quoting system exists.Shipping these stubs is better than omitting them. Without them you would get TypeError: undefined is not a function — a message that tells you nothing and sends you hunting for a typo in your own code. With them you get the actual answer: this does not exist yet, here is where to watch for it.
They will never silently start working. When intents ship, these methods gain real implementations in the same release that flips the badge on capability status.
What you can build today
The CLI scaffolder generates a solver skeleton, and it is careful about the line:
npx create-libertynet-agent my-solver --type solver -yfetchIntents()andsubmitSolution()throw. A stub that returns fake quotes is how you end up shipping fake quotes.priceIntent()is pure and testable right now. Your pricing model does not need a network that does not exist — write it, test it, and have it ready.- Discovering and verifying the nodes you would be competing with works today.
That is a real amount of useful work, done honestly, without pretending the rest is there.
No value movement
Worth repeating in the page most likely to suggest otherwise: there is no endpoint in this API that moves value. No wallet, no swap, no staking, no transfer. Intents as described here would require settlement infrastructure that does not exist and has not been reviewed.
The capability matrix, with how every claim was verified.