bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 12쪽 중 2쪽
the network actually delivers. As more agents take more actions that require more security checks, more consumption settles in $GPS and funds the operators and contributors that produce security.
**Open and verifiable network:** Through decentralized data contribution, distributed security compute, service coordination, and governance, GoPlus makes security services more transparent, scalable, and aligned with the communities they protect.
### Conclusion
The AI era requires a new security layer. As agents and applications gain the ability to execute complex actions on behalf of users, security must shift from passive detection to real-time protection before execution, and from seat-based software to a service consumed continuously by machines.
GoPlus is building that layer in both directions. As AI for Security, GoPlus applies AI to auditing, scanning, detection, and simulation through DeepScan, SafuSkill, and GoPlus Intelligence. As Security for AI, GoPlus governs agent information flows, actions, software, and on-chain interactions through Axora, SafuSkill, and the AI Agent Security API. Starting from its proven Web3 security foundation, GoPlus provides the intelligence, runtime controls, transaction security, supply-chain protection, policy enforcement, and verifiable services needed to secure high-risk digital actions across AI and Web3. $GPS is the settlement unit that ties the consumption of those services, in both directions, to the network that produces them.
# Architecture Overview
<figure><img src="https://2031722588-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3j5GFgqQKWTX18n74ll7%2Fuploads%2FsigHCDNIkZMMIsKJSAfJ%2Fimage.png?alt=media&token=a2b15317-5fef-47b0-8bc9-d17d08e83506" alt=""><figcaption><p>Architecture Overview</p></figcaption></figure>
GoPlus Security Network is the open and verifiable infrastructure behind the GoPlus security layer for the AI era. It serves two directions: AI for Security, where AI is applied to auditing, scanning, detection, and simulation, and Security for AI, where security is applied to autonomous AI systems. It provides the risk intelligence, verified security data, distributed security compute, runtime protection, and economic coordination needed to protect AI agents, Web3 users, applications, wallets, chains, and security developers.
The network started with Web3 user security, where GoPlus built real-time risk intelligence, transaction protection, malicious asset detection, and on-chain security infrastructure across multiple chains. As AI agents become a new execution environment, the same security layer expands to protect runtime behavior, tool calls, command execution, credential access, URL interaction, software supply chains, and agent-to-Web3 workflows. As the consumers of security expand from humans to machines, $GPS provides the settlement unit that connects that consumption to the network producing it.
### Core Architecture
The GoPlus Security Layer is organized around four connected infrastructure layers and one economic layer:
* **Security Intelligence**
* **Runtime Protection**
* **Security Data Layer**
* **Security Compute Layer**
* **Security Service Settlement ($GPS)**
Together, these layers create a system that can understand risk, enforce policy before execution, continuously improve through verified security data and distributed analysis, and pay the producers of security from the consumption they serve.
#### Security Intelligence
Security Intelligence is the risk engine behind GoPlus products. It evaluates assets, addresses, URLs, signatures, approvals, transactions, dApps, packages, skills, commands, prompts, and agent actions.
This layer includes token security analysis, malicious address detection, phishing site detection, transaction simulation, approval and signature analysis, dApp security information, prompt-injection detection, credential leak detection, malicious command detection, URL analysis, package risk analysis, and supply-chain risk detection.
Security Intelligence produces structured risk signals consumed by Axora, SafuSkill, DeepScan, GoPlus Web3 Security, SafeToken-related workflows, wallets, applications, and third-party integrations. It is delivered to developers through the Token Security Service and related APIs, and to autonomous systems through the AI Agent Security API.
#### Runtime Protection
Runtime Protection applies security decisions before risky actions execute.
For AI agents, Runtime Protection is represented by Axora. Axora Router governs the information flowing between applications, models, and tools; Axora AgentGuard evaluates shell commands, file access, tool calls, URLs, package content, sensitive operations, and agent-to-Web3 actions before they run; Axora Arena verifies that both hold up under attack. Axora can allow, block, warn, or require approval based on policy and risk analysis.
For Web3, Runtime Protection is represented by Security RPC, on-chain firewall, browser extension, GoPlus App, wallet integrations, dApp integrations, and other infrastructure integrations. These components analyze signatures, approvals, transactions, tokens, dApps, and addresses before users or agents submit high-risk on-chain actions.
#### Security Data Layer
The Security Data Layer collects, verifies, and distributes security data through open contribution and validation mechanisms. It supports malicious address data, phishing data, token risk data, dApp data, transaction risk data, signature risk data, agent threat data, malicious skill and package data, URL risk data, and other security intelligence inputs.
This data layer is critical because risk detection is only as strong as the security data behind it. Contributors stake $GPS to participate and earn $GPS when contributed data is verified, which aligns the supply of security data with the network that consumes it.
#### Security Compute Layer
The Security Compute Layer is the distributed network of operators that performs security analysis for the network, including AVS operators and SecHub participants. It provides the analysis capacity behind scanning, simulation, audit, and verification workloads as security-service consumption grows from human-initiated activity to high-frequency machine-originated requests.
Operators stake $GPS to provide capacity and are rewarded in $GPS from the settlement generated by the services they process. This is the supply side of the security service economy.
#### Security Service Settlement ($GPS)
$GPS runs across all of the layers above. Security services consumed through Security Intelligence and Runtime Protection are metered and settled in $GPS, and that settlement funds the rewards paid to Security Data Layer contributors and Security Compute Layer operators. GoPlus Network Governance, also coordinated through $GPS, sets the parameters of this economy. See [$GPS: The Security Service Settlement Layer](/goplus-network/usdgps/gps-security-service-economy).
### Product and Integration Components
#### Axora
Axora is the AI agent security product line of GoPlus, made up of Axora Router, Axora AgentGuard, and Axora Arena. It protects agent information flows and actions before they execute, including commands, file operations, URLs, tool calls, package scans, approval flows, transaction workflows, red-team testing, and audit timelines.
#### GoPlus Intelligence and AI Agent Security API
GoPlus Intelligence is the shared risk engine used across the GoPlus security layer. It powers token analysis, malicious address detection, phishing detection, transaction simulation, signature analysis, approval analysis, dApp risk information, and agent-oriented risk detection. The AI Agent Security API exposes it to autonomous systems through a usage-based, machine-to-machine model.
#### Web3 Security Products
GoPlus Web3 Security includes GoPlus SafuSkill, GoPlus DeepScan, SafeToken Protocol and Locker, transaction protection, Security RPC, on-chain firewall, GoPlus App, browser extension, GoPredict, and infrastructure integrations. These products protect the software agents install, token launch, contract audit and monitoring, wallet activity, user transactions, dApp interaction, and chain-level execution paths.
#### Ecosystem and Governance
GoPlus Security Network is supported by users, developers, data contributors, compute operators, security service providers, governance participants, and ecosystem partners. This ecosystem gives GoPlus a broad distribution path across chains, wallets, exchanges, DeFi protocols, launchpads, trading tools, data platforms, RPC providers, agent platforms, security partners, and infrastructure providers.
### Conclusion
GoPlus Security Network is evolving from a Web3 user security network into a broader security infrastructure for AI agents and Web3. By combining security intelligence, runtime protection, a verified security data layer, a distributed security compute layer, and $GPS settlement, GoPlus provides a scalable foundation for protecting high-risk digital actions before they execute, and for paying the producers of that protection from the consumption it serves.
# Security Data Layer
The Security Data Layer is the trusted data foundation of the GoPlus security layer for the AI era.
In the last few years, GoPlus Network has experienced exponential growth, with security data usage increasing by over 5000 times from what was recorded in 2022 and daily API calls reaching tens of millions, demonstrating high levels of ecosystem trust. As GoPlus expands from Web3 security into AI agent security, the integrity and reliability of security data becomes even more important. Agent runtime protection, transaction protection, policy enforcement, and decentralized security services all depend on accurate, timely, and verifiable risk data.
To address this need, GoPlus proposes a decentralized Security Data Contribution and Verification Layer that harnesses multi-party participation and automated verification processes. This foundational layer provides trustworthy, rich, and real-time security data for both Web3 and AI agent security. It supports the collection, verification, and utilization of token risk data, malicious address data, phishing data, dApp risk data, signature risk data, transaction risk data, prompt-injection indicators, malicious command patterns, credential leak patterns, URL risk data, package risk data, and other security intelligence inputs.
<figure><img src="https://2031722588-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F3j5GFgqQKWTX18n74ll7%2Fuploads%2Fd3uXfqU2SOsXfS9HouSv%2F%E7%99%BD%E7%9A%AE%E4%B9%A6.003.png?alt=media&token=9da8e309-36d9-41a1-9129-44504fd20b23" alt=""><figcaption><p>Security Data Layer</p></figcaption></figure>
### Data Contribution
Security data contributors form the foundation of the entire system. They provide valuable information about potential security risks and threats through various channels, including but not limited to:
**End-users:** Users can report security issues they encounter, such as suspicious scam activities, phishing attempts, malicious websites, wallet drainers, rug pulls, and unsafe agent actions.
**Security researchers:** Professional security researchers can contribute findings on Web3 risks, AI agent threats, malicious packages, prompt-injection payloads, credential leakage patterns, and other in-depth security insights.
**Third-party security companies & organizations:** Specialized security firms can offer comprehensive threat intelligence and risk assessment reports.
Through incentivization and recognition mechanisms, GoPlus encourages broad participation to establish a comprehensive and diverse security database.
### Data Verification
To ensure the credibility and accuracy of the contributed data, we implement a multi-tiered decentralized verification mechanism. The Security Data Verification system consists of a Primary verification process and a Secondary verification process, working in tandem to validate the security data.
#### **Primary verification**
The Primary verification employs a multi-faceted approach to data verification, incorporating trusted third-party entities and automated computational methods:
**Third-Party Verification Nodes:** Reputable entities operate verification nodes that leverage their expertise and resources to assess the veracity of user-contributed information.
**Computational Verification Nodes:** Automated computational methods are used to verify specific types of security data through security scanning, advanced algorithms, and AI techniques.
**Auditors:** Independent auditors oversee the verification process, ensuring compliance with established protocols and maintaining the integrity of the system.
#### **Secondary verification**
The Secondary verification is triggered when disputes arise in the Primary verification. It is composed of highly specialized security teams and institutions that focus on resolving controversies in Primary verification:
**Elite Security Teams:** Renowned security teams with expertise in Web3 security, AI security, application security, and threat intelligence are enlisted to investigate and resolve complex disputes.
**Institutional Arbitrators:** Respected institutions, such as respected university labs and Web3 industry leaders, act as impartial arbitrators to settle disagreements and provide final verdicts.
The Secondary verification ensures that any contentious issues are thoroughly examined and resolved by the most qualified experts in the field.
By seamlessly integrating security data contributors and the multi-tiered verification mechanism, GoPlus creates a robust and resilient decentralized security data ecosystem. This approach enhances the diversity, professionalism, and accuracy of risk data while strengthening the risk models that protect users, AI agents, wallets, applications, and chains. By working together, the ecosystem can build a shared data foundation for protecting high-risk digital actions before they execute.
# Types of Risk Data
GoPlus has identified and prioritized a range of critical security data types. These data types serve as the backbone of our decentralized Security Data Contribution and Verification Layer, providing comprehensive insights into potential security risks and enabling mitigation strategies. Here's an overview of these data types:
#### **Token Security Data**
This category encompasses analyses of token contracts, highlighting potential risk assessments, and token holder distribution analyses. Additionally, we have introduced an open source [Token Risk Classification](/goplus-network/architecture/security-data-layer/token-risk-classification) standard, a framework designed to categorize the various risks associated with tokens. Token security data plays a vital role in offering stakeholders a detailed understanding of the security aspects of token projects, aiding in the identification and mitigation of associated risks. This classification standard further enhances our ability to assess and communicate the nuances of token-related risks effectively.
#### **Malicious Address Data**
Malicious address data includes known blockchain addresses associated with scams, phishing, hacking and other fraudulent activities. By identifying and warning users about these addresses, this data type is crucial in preventing interaction with these malicious addresses and enhancing user security.
#### **NFT Security Data**
This category encompasses analyses of NFT contracts, highlighting potential risk assessments, and token holder distribution, NFT information analyses. NFT security data plays a vital role in offering stakeholders a detailed understanding of the security aspects of NFT projects, aiding in the identification and mitigation of associated risks.
#### Approval Risk Data
Approval risk data primarily focuses on potentially hazardous contracts that require user authorization, including contracts that have been compromised in hacker attacks as well as malicious contracts. When users authorize their assets to these contracts, they may face the risk of asset loss. This type of security data is crucial in helping users identify and revoke permissions to dangerous contracts, thereby preventing the authorization of their assets to these risky entities. Approval risk data serves as a vital tool in safeguarding user assets against unauthorized access and potential misuse by highlighting the risks associated with certain contract authorizations.
#### **dApp Security Data**
This category comprises security audit reports of smart contracts, known vulnerability lists, and community safety feedback. dApp security information provides a comprehensive safety assessment for dApp users, helping them avoid interactions with insecure dApps.
#### **Specific Malicious Signature Features Data**
Targeting anomalies and potential risks in blockchain transaction signatures, such as unauthorized transactions or suspicious contract calls, this data helps identify and prevent malicious activities, enhancing transaction security.
#### **Phishing Site Data**
Phishing site data involves characteristics of known phishing sites and user feedback, aimed at identifying potential phishing attacks. This data is vital in preventing users from accessing malicious websites and protecting them from data or asset theft.
### Conclusion
Together, these security data types form the core of our decentralized data contribution system. By integrating and analyzing this data, the network can more effectively identify and respond to security threats, ensuring the safety of users and their assets. This collective effort lays a solid foundation for the future of digital interactions, empowering all participants to navigate the Web3 world with confidence and security. Furthermore, we plan to enrich and expand the variety of security data categories through governance and voting mechanisms in the future. This approach will enhance the diversity and coverage of our security data, strengthening the overall robustness of our security data ecosystem.
# Token Risk Classification
Token Risk Classification(TRC) aims at identifying and cataloging scams like honeypots, and intentional backdoors that may be present in token smart contracts within the web3 ecosystem. This classification serves as:
* **A Shield against Malicious Smart Contracts:** By showcasing a defined list of malicious token contract patterns, it empowers users and project teams to recognize and steer clear of contracts with hidden intents, thereby ensuring safer interactions within the decentralized space.
* **A Testing Ground for Developers:** With a clear classification of malicious patterns and real-world examples, developers creating tools to detect these malicious token smart contracts can effectively evaluate their systems against a standardized classification.
* **A Catalyst for Research:** By clarifying the deceitful practices adopted in token smart contracts, we hope to drive more research towards crypto user safety, encouraging the community to devise strategies that deter such behaviors.
* **An Educational Asset:** This Github repository stands as an initiative to amplify awareness, serving as an informational storage hub, shedding light on potential contract pitfalls and deceitful patterns to the advantage of the community.
For more details, please visit our [TRC website](https://cryptousersecurity.github.io/token-risk-classification/).
### Contribute <a href="#contribute" id="contribute"></a>
Maintaining the relevance and comprehensiveness of this Github repository is a joint endeavor. We hope for and welcome community contributions. For details on how to contribute, kindly refer to our [Contribution Guidelines](https://github.com/cryptousersecurity/token-risk-classification).
<br>
# Axora
## Overview
프로젝트로 돌아가기