hackquest logo
  • 学习
  • 课程探索
  • 实战
  • 探索黑客松
  • 往期项目
  • 社区
  • 论坛
  • 全球活动
  • 我的社区
  • 共学营
  • 组织中心
  • 更多
    • 词汇表
    • 博客
    • 水龙头
    • 工作机会
    • 媒体工具包
    • 创始人
hackquest logo

Sumplus

AI agent execution infrastructure for Robinhood Chain and Arbitrum, helping agents access models, call tools, monitor markets, and execute onchain actions with better control.

视频

项目图片 1

技术栈

Web3
React
Java
Move
Solidity

描述

Sumplus Cashier is a control and verification layer for AI agent spending on Arbitrum and Robinhood Chain.

As AI agents begin to call tools, APIs, models, and services on their own, a normal gateway only shows the bill after money is spent. Cashier adds control before execution and proof after execution.

Before each paid call, Cashier checks policy rules such as per-call ceiling, session cap, host allowlist, and allowed actions. If the call is allowed, it proceeds to the upstream service. If it is refused, the refusal is also recorded with a clear reason.

After each decision, Cashier creates a hash-chained receipt. Anyone can recompute the receipt chain and verify whether the spending record has been modified, without trusting Sumplus as the operator.

For this buildathon, Cashier is positioned for agentic finance on Arbitrum and Robinhood Chain, where AI agents may need to pay for models, data, APIs, trading signals, and execution services under clear spending limits.

Live demo:
https://sumplus-cashier-production.up.railway.app

GitHub:
https://github.com/sumplus-real/sumplus-cashier

本次黑客松进展

Sumplus Cashier has a working live demo and public GitHub repository.

Current progress includes:

  • Live deployed demo on Railway

  • Policy checks before agent-paid calls

  • Refused calls with human-readable reasons

  • Hash-chained receipt ledger

  • Receipt verification page

  • Negative control test showing edited receipts are rejected

  • Integration with Arsenal capability data

  • Integration with SafeRouter attestation and verification flow

  • Local run instructions and verification scripts

The current demo is read-only and does not move real funds. It demonstrates how AI agent spending can be controlled and verified before being extended to real payment and execution flows on Arbitrum / Robinhood Chain.

融资状态

Sumplus is currently in an early product and ecosystem growth stage.

The team is focused on building trusted execution infrastructure for AI agents through SafeRouter, Arsenal, Maria, and Cashier.

Cashier extends this stack by adding policy-based spending control and verifiable receipts for agent payments, especially for agentic finance, tokenized assets, AI-driven trading, and onchain service payments on ecosystems such as Arbitrum and Robinhood Chain.

Current traction includes:

  • 48K+ X / Twitter followers

  • 51K+ global community users

  • 5000+ core users

  • 300+ ecosystem partners

  • Arsenal with 166 live Skills

  • 2 hackathon recognitions: Solana × Metaplex Hackathon 2nd Place and Mantle Turing Test Hackathon 2026 Project Deployment Award

Sumplus is open to ecosystem partnerships, grants, strategic support, and potential investment discussions.

队长
OOrangecat Huang
项目链接
赛道
AI