Today, we’re launching Arsenal — a specialized marketplace of skills, tools, and APIs built for on-chain AI agents. It’s the piece that turns an agent that reasons about capital into one that can act on it, safely and composably, on Sui.

What Is Arsenal?

If Maria (our execution agent) is the brain and security guard of the Sumplus stack, Arsenal is the toolbox.

Arsenal breaks complex DeFi operations — swapping, yield farming, providing liquidity — into standardized, callable modules. Any AI agent can discover, equip, and execute these skills without hardcoding every protocol interaction.

The vision is simple: financial capabilities should be secure, callable, and easy to compose.

Key Features

1. Modular “Plug-and-Play” Skills

Arsenal packages DeFi operations into standardized modules. Developers integrate once and get access to a growing library of financial primitives — no need to rebuild protocol-level integrations from scratch.

2. 40+ Skills, Live on Sui

Arsenal launches with nearly 40 skills already live on the Sui network. Our flagship is Yield — a callable, optimized path for agents to access yield opportunities through Sumplus in a single API call, best execution, gas handled. More skills are being added continuously.

3. Secure Execution via Maria

Every Arsenal skill runs inside Maria’s trust layer. That means user-authorized, policy-enforced execution. Funds never leave user custody. Agents act within defined permission boundaries.

Connecting Your Agent to Arsenal

Arsenal exposes a standard REST API. Whether your agent is built on LangChain, AutoGen, a custom framework, or raw inference calls — integration takes minutes.

The flow looks like this:

1. Register
2. Authenticate via Maria
3. Call skills
4. Receive signed execution

When you register an agent, you define exactly which skills it can use and set spending limits upfront. Each agent identity is scoped — so a yield-optimization agent has no ability to call swap functions unless you explicitly grant it.

Code Snippets

Registering an Agent:

JSON

POST /v1/agents/register
{
  "agent_name": "my-yield-optimizer",
  "allowed_skills": ["yield"],
  "spending_limit_usd": 500,
  "auth_policy": "user_confirm_above_100"
}

Executing a Skill:

JSON

POST /v1/skills/yield/execute
{
  "asset": "USDC",
  "amount": 1000,
  "strategy": "auto_optimize",
  "slippage_bps": 30
}

Arsenal returns the transaction hash, achieved APY, and position ID. Gas is abstracted — your agent doesn’t need to hold SUI.

Available Skills

• Yield (✅ Live): Deposit into optimized yield positions on Sui.
• Swap (🔜 Coming soon): Cross-DEX asset swaps with best-execution routing.
• Liquidity (🔜 Coming soon): Programmatic LP position management.
• Portfolio read (🧪 Public beta): Read-only portfolio state, no signing required.

What’s Next

Nearly 40 skills live is a start, not a ceiling.

The near-term roadmap focuses on three directions:

• Deepening the Skill Library: More DeFi primitives and protocol integrations to cover the full range of on-chain actions.
• Skill SDK: We are opening Arsenal to external developers. A new SDK will let anyone package and publish their own modules to the marketplace.
• Composability: Enabling agents to chain skills together to build higher-order strategies and operate with real financial autonomy.

The broader ambition is to make Arsenal the default action layer for any AI agent that touches capital. Not just on Sui — but as the model for what agent-native financial infrastructure should look like across the industry.

We’re building the toolbox. The agents will do the rest.

Sumplus is a composable financial stack for AI agents on Sui. Our products include Maria (execution agent), Arsenal (skill marketplace), and Yield. Follow us at @sumplusreal.

In October 2025, Tempo closed a $500 million Series A at a $5 billion valuation, led by Thrive Capital and Greenoaks, with Stripe and Paradigm as founding backers.

The round is notable not only for its size but for what it signals: Stripe, which stepped back from crypto payments in 2018 citing volatility and unreliability, chose to re-enter the space through a purpose-built, stablecoin-native payment chain.

SumPlus Arsenal has integrated Tempo. This post explains the reasoning behind that decision — and what we believe it means for the agents and developers building on our platform.

SumPlus Arsenal: The Context

SumPlus is a composable financial stack for AI agents. Its core components address distinct layers of agent finance: Maria handles secure onchain execution; Arsenal is a cross-chain skills marketplace where agents discover, access, and compose financial capabilities.

The premise underlying both is straightforward.

As AI agents become more capable, the limiting factor in their usefulness is increasingly not reasoning or planning — it is the ability to act financially. An agent that can analyze a market opportunity but cannot execute a trade, pay for a data feed, or settle a service invoice is an agent whose capabilities are artificially constrained by infrastructure that was never designed with it in mind.

Arsenal was built to remove that constraint at the capability layer.

But a skills marketplace is only as fluid as its payment infrastructure. If accessing a skill requires pre-registration, API key management, or human approval of each transaction, the composability that makes a marketplace valuable collapses into something closer to a directory. The payment layer has to be as programmable as the skills themselves.

A Shared Starting Point

Most financial infrastructure — including most blockchain infrastructure — was designed around human transaction patterns.

Accounts are owned by people. Authentication assumes a person is present. Settlement timelines are calibrated to human workflows. Gas management assumes a user who can monitor balances and top up when needed. Even in decentralized systems, the implicit user is still a person making deliberate choices.

This assumption breaks down for autonomous agents.

An agent does not have an attention span to manage gas. It cannot complete a sign-up flow before calling a service. It cannot wait for a settlement window that outlasts its execution context. It needs payment to work the way API calls work: reliably, instantly, and without out-of-band setup.

Tempo was built against this specification. Its Machine Payments Protocol (MPP), co-authored with Stripe, formalizes a model in which payment is a native property of an HTTP request rather than a separate process layered on top of one:

1. A client requests a resource
2. Receives a 402 Payment Required response with a payment challenge
3. Settles the challenge
4. Retries — all within a single request cycle

No redirects. No webhooks. No prior relationship between client and server required.

The session model addresses the economics of high-frequency agent workflows: rather than settling each micropayment on-chain, an agent opens a session by depositing funds into escrow and issues signed off-chain vouchers as it consumes resources. Periodic on-chain settlement aggregates these into a single transaction, making per-call billing economically rational at scale.

Both SumPlus and Tempo, working from different directions, reached the same conclusion: serving autonomous agents well requires infrastructure designed for them from the ground up, not adapted from systems built for people. That alignment in first principles is what makes this integration coherent rather than merely convenient.

What Is Live

The integration is not aspirational.

Arsenal has already integrated Uniswap deployed on Tempo, enabling agents to access onchain liquidity and execute token swaps directly within Arsenal’s skill framework, with settlement handled by Tempo’s payment infrastructure.

An agent can invoke a liquidity skill on Arsenal, execute a swap via Uniswap on Tempo, and settle in stablecoins — within a single autonomous workflow, without human intervention.

With Arsenal’s MPP integration live, every skill in the marketplace is natively payable by any agent holding a Tempo-compatible wallet. An agent running a Maria-orchestrated strategy can call an Arsenal skill, pay via MPP, receive the result, and incorporate it into its next decision — all within a continuous autonomous loop.

For skill developers, this changes the economic model in a meaningful way. A skill listed on Arsenal is now accessible to any MPP-capable agent on the open internet, not just agents provisioned through SumPlus. Settlement is immediate. Discovery is open. The Payments Directory, which already catalogs over 100 MPP-compatible services, extends Arsenal’s reach further still.

The Broader Significance

Taken together, the SumPlus stack and Tempo’s payment infrastructure form a reasonably complete picture of what agent-native finance requires.

Maria provides the execution environment. Arsenal provides the capability layer. Tempo’s MPP provides the settlement rails. Each component addresses a distinct constraint that would otherwise limit what agents can do autonomously.

The significance of this is not that it produces a more efficient version of existing financial infrastructure. It is that it produces a different kind of financial infrastructure — one in which agents are first-class participants rather than edge cases accommodated within systems designed for people.

This is a gradual process, not a sudden transition. Most financial activity today still flows through human-operated systems, and that will remain true for some time. But the direction is clear: as agents become more capable and more widely deployed, the demand for infrastructure that meets them on their own terms will grow.

The work being done now — by Tempo on the payment layer, by SumPlus on the financial capability layer — is the foundation on which that future is being built.

SumPlus is a composable financial stack for AI agents, built on Sui. Arsenal is SumPlus’s cross-chain skills marketplace, with 40+ live Skills spanning Sui, Solana, Raydium, Jupiter, and more.

Follow us: @sumplusreal

We stand at the precipice of a tectonic shift in finance: AI is becoming the reasoning engine, and Web3 is the settlement layer. In this new paradigm, machines don’t just suggest strategies; they execute them. But between the logic of an LLM and the finality of the blockchain lies a critical, often ignored gap. In a world where “Code is Law,” there is no undo button.

One hallucinated contract, one misinterpreted prompt, or one interaction with a malicious pool, and your capital is gone. Instantly. Irreversibly.

Sumplus was built to close that gap.

1. Why the Brain Shouldn’t Hold the Keys

The fundamental flaw in today’s AI agents isn’t their intelligence — it’s their architecture.

When the same system that reasons also signs the transaction, a single point of compromise leads to a total loss. Hallucinations, prompt injections, and adversarial inputs aren’t just bugs; they are massive attack surfaces.

“If the AI is the pilot and the blockchain is the engine, we are currently flying without a flight control system.”

Sumplus decouples these two functions. Our execution agent, Maria, sits between AI reasoning and on-chain action as a dedicated gatekeeper. Before any transaction reaches the chain, Maria performs a singular, vital check: Does this action actually match the original intent?

This is IAA — Intent-Action Alignment.

IAA isn’t a simple policy list or a rigid rule engine. It is a provable, deterministic question: Did the agent do exactly what it meant to do? If the answer is anything less than a definitive “Yes,” the transaction is killed before it ever hits the ledger.

2. Constraining the Action Space

You cannot align actions that you cannot define. This is where Arsenal comes in.

Through Arsenal, our skill marketplace, we have crystallized complex DeFi protocols into standardized, auditable “Skills.” Agents operating within the Sumplus ecosystem don’t guess at what’s possible — they operate within a pre-vetted vocabulary of financial actions.

By constraining the action space from the start, we make IAA verification not just possible, but mathematically tractable and reliable.

3. Verification You Can Trust (Not Just Feel)

In decentralized finance, “feeling” reliable isn’t enough.

Every IAA check Maria runs is recorded on-chain — making the entire process auditable, transparent, and verifiable by anyone. We aren’t asking users to trust the AI’s “black box.” We are providing the infrastructure to verify the output.

The Sumplus Stack: A New Foundation

• Arsenal: Defines the boundaries of safe, executable financial actions.
• Maria: The guardian that runs IAA before every on-chain transaction.
• Yield: Our first battle-tested strategy, proving that high-performance DeFi can coexist with rigorous safety.

Agentic finance is no longer a future concept; it is arriving. IAA is how we make sure it’s safe enough to inhabit.

Web: sumplus.xyz

X:https://x.com/SumPlusReal

— How OpenClaw Supercharges the Leap Toward Autonomous Agentic Finance

In the 2026 Web3 ecosystem, Decentralized Finance (DeFi) is undergoing a qualitative transformation from “liquidity mining” to “intelligent execution.” The Agentic Financing Stack proposed by SumPlus, by integrating the OpenClaw general-purpose agent framework, is more than just a financial tool; it is becoming a “composable living organism” capable of autonomously managing, growing, and protecting user wealth.

1. Solving the “Last Mile” of the AI Ecosystem: From Intention to Execution

The current AI ecosystem (ECO) faces a severe “fracture”: Large models can provide perfect investment advice but cannot press the “Confirm Transaction” button for you. SumPlus is dedicated to bridging this final execution gap.

• The End of Intention Friction: Traditional DeFi requires users to understand complex slippage, Gas fees, and cross-chain bridges. SumPlus encapsulates these complexities within the underlying layer.
• Agentic Financing Stack: We have built a financing stack specifically designed for agents, enabling AI to interact directly with smart contracts.
• Financial-Grade “Certainty”: Through the MCP (Model Context Protocol), we translate non-deterministic AI language into deterministic on-chain instructions, ensuring that “last mile” execution is both secure and precise.

2. OpenClaw & SumPlus: The “Muscle” Behind the Financial Brain

If SumPlus is the brain, then OpenClaw is its hands and eyes. As a general-purpose agent framework, OpenClaw injects unprecedented interoperability into SumPlus.

• Infrastructure Layer: SumPlus has established an MCP-compatible platform. This is not just a collection of APIs, but a financial Skill market. Developers can contribute specific “financial scrapers” or “execution” plugins (Claws) and earn tokenized yields by providing these services.
• Protocol Layer — Agent-Native DEX: The SumPlus protocol layer is an agent-native DEX. It abstracts complex liquidity pool logic into simple, agent-ready primitives. OpenClaw can utilize these primitives to complete multi-path rent-seeking and risk hedging within milliseconds.
• Application Layer — SumPlus Maria: As the core application, SumPlus Maria is the world’s first DeFi-native agent.
• EvoMap (Strategy Evolution Graph): Maria no longer runs rigid scripts. Instead, she autonomously evolves trading strategies through EvoMap, correcting logic in real-time based on market sentiment and on-chain data.
• A2A (Agent-to-Agent) Collaboration: With OpenClaw, Maria can engage in “agent-to-agent dialogues” with other general-purpose agents (such as your personal work assistant), enabling a frictionless financial experience — like “automatically allocating payroll to high-yield strategies”.

3. Web4 Compatibility: The Era of Autonomous Symbiosis

The ultimate vision of SumPlus is to exist as a native component of Web4. In the Web4 era, the internet is no longer a collection of web pages, but an intelligent mesh composed of countless collaborative autonomous agents.

1. Extreme Markets Expose the Limits of Human-Centric Finance

Recent market drawdowns have once again demonstrated a hard truth of modern crypto markets.

BTC and ETH experienced rapid declines, triggering waves of forced liquidations across chains and exchanges. Even well-capitalized “whale” accounts — many of which had previously accumulated significant unrealized gains — were liquidated within hours.

This was not due to a lack of experience or capital.

It was the result of reaction speed mismatch: markets evolved faster than human-operated risk systems could respond.

As financial systems become more complex, highly leveraged, and reflexive, human-in-the-loop finance is increasingly misaligned with market reality.

2. From Risk Models to Risk Agents

Traditional financial risk management is model-centric. Parameters are predefined, thresholds are fixed, and responses are reactive. Even sophisticated quantitative systems rely on factor sets designed in advance.

SumPlus is built on a different paradigm: agentic financing.

In the SumPlus framework, risk is not managed by static models, but by autonomous AI agents capable of perception, reasoning, and action. These agents do not simply observe price movements — they continuously interpret market structure, liquidity dynamics, leverage concentration, and narrative formation.

Rather than asking “Has a risk threshold been crossed?”, SumPlus agents ask:

“Is the system becoming fragile?”

This shift allows risk to be identified before it materializes as liquidation events.

3. Risk as a Living, Adaptive Process

Agentic financing treats risk as a dynamic process rather than a static variable.

SumPlus agents continuously ingest signals from:

· On-chain flows and position changes

· Centralized exchange order books and funding data

· News, macro events, and policy signals

· Social media and narrative consensus formation

Unlike traditional quant systems, where factor importance is fixed, SumPlus agents dynamically determine which signals matter most at each point in time.

For example, when large research-driven funds or well-known market participants adjust their positions, AI agents detect not only the on-chain impact, but also the amplification effect created by social consensus and narrative propagation. These signals are automatically elevated in priority and reflected in risk assessments and portfolio guidance.

4. Front-Loading Risk Through Autonomous Reasoning

The core objective of SumPlus is risk front-loading.

Instead of responding after volatility spikes, SumPlus agents reason about early indicators of instability — including leverage clustering, liquidity thinning, and sentiment convergence — and adjust recommendations accordingly.

This may result in:

· Earlier exposure reduction

· Portfolio rebalancing suggestions

· Risk alerts tied to structural fragility, not price alone

Importantly, agents operate continuously. Risk management no longer depends on manual monitoring or periodic reviews.

5. From Human Decision Support to Agent-Operable Finance

Agentic financing is not about replacing humans. It is about changing the interface between humans and capital.

SumPlus converts financial infrastructure into AI-callable primitives, allowing agents to:

· Monitor positions persistently

· Analyze portfolio risk holistically

· Propose and simulate adjustments

· Trigger alerts and execution logic

For users, complexity is abstracted away.

What they interact with is not raw data, but agent-generated financial intelligence.

6. Infrastructure for an Autonomous Financial Future

As markets continue to expernce extreme volatility, financial systems must evolve beyond reactive frameworks.

SumPlus is building the foundational layer for agentic financing — where autonomous agents manage risk as a first-class function, continuously and adaptively.

The goal is not to predict markets or eliminate uncertainty.

It is to align financial systems with the speed and complexity of modern markets.

In an environment where downside is not an exception but a constant, agentic risk management becomes essential infrastructure.

SumPlus is built for that future.

🌐 Website → https://sumplus.xyz

🐦 X → https://x.com/SumPlusReal

💬 Telegram → https://t.me/sumplus_official

📚Deck → https://www.papermark.com/view/cmifrvdr90001l104vltn34ov

Stablecoins have quietly become the foundation of digital finance.

They power trades, payrolls, and payments — silently moving trillions across the internet.

In just five years, the total market cap of stablecoins has ballooned from $50 billion to over $250 billion, with USDT and USDC together representing nearly 90% of supply.

They are the liquidity layer of the crypto economy — yet they remain, ironically, liquidity without life.

They don’t earn, adapt, or evolve.

They simply exist.

In a world where everything is programmable, the idea that “money should sit still” feels outdated.

At Sumplus, we believe that every stablecoin should work for its holder — continuously, intelligently, and safely.

That belief gave birth to our mission:

to turn stablecoins into living assets.

While DeFi exploded with innovation, one obvious gap remained: there is still no native yield layer for stablecoins that can serve both institutions and individuals.

Today, less than 6% of all stablecoins are earning yield. The rest are idle — sitting on exchanges, in wallets, or parked in treasuries.

Why? Because yield in crypto today is fragmented, opaque, and often short-lived.

Most protocols are built for “DeFi natives,” not everyday users.

They offer 20-click deposit flows, volatile APYs, and complex token mechanics.

It’s finance for experts, not for everyone.

Sumplus aims to change that — by reimagining what a stablecoin can do once it starts to earn.

The core idea is simple but powerful:

two tokens that work as one system.

• aUSD — a 1:1 dollar-pegged stablecoin minted from deposits like USDT, USDC, USD1, or crypto collateral such as BTC, ETH, and SUI.
• saUSD — a yield-bearing representation of aUSD that continuously appreciates as yield accrues.

This design decouples stability and yield.

aUSD ensures seamless payments, transfers, and settlements — the utility of traditional stablecoins.

saUSD represents ownership of productive capital — an asset that compounds automatically without staking, locking, or manual claims.

You can pay with it, transfer it, or simply hold it — your yield never stops.

We think of aUSD as the Layer-2 stablecoin, or M2 of money — built on top of M1 assets like USDT and USDC, extending their utility from static value to dynamic yield.

The process is effortless:

Deposit → Mint → Earn → Redeem.

When you deposit assets, you mint aUSD.

That aUSD is allocated into AI-managed yield pools — diversified across multiple strategies such as:

• Delta-neutral hedging: capturing perpetual swap funding spreads while neutralizing market exposure;
• RWA yield: tokenized short-term Treasuries and corporate bonds;
• Crypto-native yield: funding arbitrage and liquid staking derivatives (LSDs) on Sui and other ecosystems.

Each strategy is monitored and dynamically balanced by the Sumplus AI Engine, which constantly tracks volatility, liquidity, and return differentials.

The system rebalances positions before risk surfaces, ensuring that every aUSD dollar is working efficiently while staying protected.

To the user, this feels like a simple deposit account.

Behind the scenes, it’s a multi-layered yield orchestration engine powered by AI.

The AI layer is where Sumplus truly differentiates itself.

It’s not just automation — it’s intelligence.

The AI Risk Engine monitors thousands of variables in real time, from CEX funding rates and on-chain liquidity to macroeconomic indicators like treasury yields.

It aggregates these signals into a unified knowledge graph, which feeds our risk models and reinforcement learning systems.

This allows the engine to do three critical things:

• Predict volatility before it happens;
• Optimize yield allocations across multiple strategies;
• Pause or rebalance pools automatically when anomalies occur.

Sumplus becomes a living, learning organism — constantly adjusting its own capital flows.

It’s asset management redefined: from human intuition to machine intelligence.

Security, of course, is non-negotiable.

Every component of Sumplus is designed to institutional standards.

Our custody infrastructure leverages Fireblocks, Copper, and CEFFU, combined with Off-Exchange Settlement (OES) — separating trading execution from asset storage.

All smart contracts are governed via Gnosis Safe multisig (5/7), with triple independent audits underway.

If anomalies are detected, the system’s Gatekeeper Module can trigger global pauses or staged withdrawals, ensuring user protection even in extreme conditions.

This is how we define stability — not as a static peg, but as a dynamic equilibrium between risk and return.

Sumplus isn’t built in isolation.

We’re part of a growing network of partners who share a vision for safer, smarter yield.

Through integrations with VISA/MasterCard payment rails and exchange partners, we’re creating a system where yield-bearing stablecoins can move, pay, and settle just like traditional money — but better.

Because yield isn’t valuable if it can’t flow.

Ultimately, our mission is not to replace existing stablecoins, but to complete their evolution.

We envision a world where the line between holding and earning disappears; where users can pay with yield-bearing money that grows in value as they sleep.

In that world, yield is not a privilege of institutions — it’s a right of every saver.

Sumplus is building the rails for that reality.

The name Sumplus comes from two ideas:

“Sum,” the total of all value, and “Plus,” the act of adding beyond it.

Together, they represent what we stand for — helping users not just hold wealth, but let it expand.

In the next chapter of DeFi, the future of liquidity is not just deep — it’s alive.

Sumplus is here to make that happen.