The real cost of an agent payment layer isn't the fee. It's the hours a developer loses when a payment fails and nobody can say why.
Kite is a good project to run that test on. Mainnet went live in April with Agent Passport and 90+ service providers already integrated, plus x402 and MCP support, a kpass CLI, and a testnet. For builders, the fully supported path today is MCP: agents like Cursor and Claude Desktop connect to Kite, and payments run inside a session with a spending limit.
What I like: the docs list real failure cases like expired sessions and insufficient budget, with fixes. What I'd watch: sessions expire, so "run it again tomorrow" means re-authenticating, and two of the three developer modes, the ones for programmatic setup, are still marked "coming soon." #KiteBBW
None of that is a red flag this early. It's just the gap between strong architecture and a smooth build.
So my test is simple: can an ordinary developer build an agent that pays for an API in one weekend, then run it again the next day without guessing?
If yes, developer experience becomes Kite's real edge. If not, great architecture stays a diagram.
Where do builders get stuck first?
📊 Building an agent that pays per API call. What makes you quit first?
A private instruction that only works on one chain is useful, but limited. That's the gap I want Nillion to close after Dusk's Ethereum mainnet launch on 28 September.
Here's how Covenants work today. You attach an instruction to a condition, it stays encrypted while it waits, and it opens when that condition is met. Good start, but it's stuck on one chain.
That's why Darkfall and Skyglow caught my eye on Nillion's seven-phase roadmap.
Darkfall is scheduled for November 2026 to February 2027. The idea is to bring everything Dusk does to multiple EVM chains through one network, with chain-specific nodes and tooling so operators and developers can work natively on each chain. @Nillion
Skyglow goes a step further. The roadmap describes a condition being verified on one chain while the instruction settles on another, with cryptographic proofs checking the reveal. It also targets support beyond EVM chains. #NIL
What I care about most is simple: the released action should follow its condition without adding a new party to trust. Connecting chains is the easy part. Holding that guarantee through the handoff is the tricky bit, and I suspect it's where most cross-chain designs struggle. Get it right and private instructions stop feeling so niche.
To be clear, Darkfall and Skyglow are roadmap goals, not live features.
Which roadmap step should Nillion deliver first? $NIL
went into @Nillion 's Encrypted Markets thinking trading privacy was mostly about people seeing your wallet activity. Then I started looking at Covenants and realised there's another problem hiding in plain sight.
Your strategy can be exposed before your trade even happens.
Think about it.
When you place an order, others see ur entry price or liquidation level. They don't need to know who you are to understand your plan. Once they see it, they can react before your trade goes through.
Here's where Covenants get interesting.
Instead of publishing an instruction in readable form, you seal it to a predefined condition. The encrypted instruction stays unreadable while Nillion's node network waits for that condition. When it triggers, nodes release the necessary cryptographic shares, allowing the instruction to be reconstructed and resolved on Ethereum.
Your strategy can be prepared in advance without broadcasting its details to everyone watching the chain.
Hmm. The part I keep coming back to is that this isn't simply about hiding a transaction. It's about changing what other market participants can learn before execution.
Imagine setting a conditional ETH buy without exposing your intended entry beforehand, or preparing a stop-loss that stays sealed until its trigger arrives.
Of course, confidentiality doesn't guarantee better execution, eliminate every form of MEV, or prevent every implementation flaw. The conditions and execution mechanism still matter.
But the idea is worth paying attention to.
We've spent years making financial activity transparent enough to verify, sometimes at the cost of revealing strategies traders would rather keep private.
so what do you like most about covenants??? #NIL $NIL