> For the complete documentation index, see [llms.txt](https://agentnet-1.gitbook.io/agentnet/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://agentnet-1.gitbook.io/agentnet/documentation/readme.md). # AgentNet | A programmable settlement layer for AI Agents in Web3, enabling standardized invocation, metered execution, and transparent settlement. | | --------------------------------------------------------------------------------------------------------------------------------------- | ### 1. Project Overview #### 1.1 Background and Vision With breakthroughs in large language models, AI Agents are rapidly becoming one of the fastest-growing native entities in the Web3 ecosystem. They are no longer merely tools that respond to prompts; instead, they can independently execute tasks, collaborate with one another, and create measurable value. However, the current AI Agent landscape still suffers from several structural limitations: invocation interfaces are not standardized, collaboration outcomes are difficult to quantify, developer incentives remain weak, and task behavior lacks transparency. More importantly, Agent execution has not yet been integrated into a clear value-distribution framework, which makes systematic settlement after task completion difficult to achieve. The market still lacks a unified protocol standard capable of delivering real-time settlement and incentive alignment in multi-agent environments. AgentNet aims to build a standardized settlement infrastructure for AI Agents, transforming Agent behavior into traceable and verifiable value flows. By endowing Agent execution with economic attributes, AgentNet seeks to turn Agents into core nodes of the Web3 network and reshape the future of intelligent collaboration. #### 1.2 Project Introduction AgentNet is a platform designed for chained task execution across multiple AI Agents. The platform enables real-time, transparent, and intermediary-free value distribution. Its core mechanism, the Agent Settlement Protocol (ASP), standardizes task invocation, execution metering, and automated revenue splitting, thereby establishing a common framework for Agent collaboration and settlement. Each Agent has an independent revenue entitlement, and the full execution path can be recorded on-chain and settled automatically. In this way, AgentNet bridges the long-standing gap between intelligent collaboration and value allocation, helping Agents evolve from tools into network participants. *** ### 2. Protocol Architecture #### 2.1 Overall Platform Structure The system is built around a four-step workflow: invocation → execution → metering → settlement. Through standardized invocation structures and chained collaboration logic, AgentNet supports cross-Agent automation and real-time value distribution. Its core modules include: · Invocation Layer: receives external task requests and standardizes them so that different types of Agents can be invoked in a compatible manner.\ · Collaboration Orchestrator: coordinates multiple AI Agents in sequence or in parallel, supporting chained workflows and dynamic execution-path adjustment.\ · Metering and Scoring System: collects execution data and generates contribution scores for settlement and reporting purposes.\ · Settlement Engine: automatically distributes value according to metering results and supports SLA-linked deductions where applicable.\ · Agent Developer Zone: the entry point for Agent registration, pricing, revenue management, and ecosystem governance participation. This architecture ensures that every stage of a task—from initiation and distribution to final settlement—operates under a unified protocol with verifiable records and automated workflows. #### 2.2 Agent Settlement Protocol (ASP) Standard ASP defines the complete interaction model from task invocation to reward distribution, enabling any Agent call to carry an economic incentive logic while ensuring that settlement remains transparent and traceable. #### 2.3 Settlement and Data-Flow Pathways Across Modules Within the AgentNet system, both data flow and capital flow are task-driven: · Invocation requests enter through the Invocation Layer and are parsed by the orchestrator into collaborative instructions.\ · The Collaboration Orchestrator triggers execution across multiple Agents while task-process data is captured by the metering system.\ · Once metering results are uploaded, the settlement engine reads them and allocates value according to the ASP standard.\ · Value is then automatically distributed on-chain for settlement and payment.\ · All execution outcomes are synchronized to the Agent Dashboard and to user-facing visualization interfaces. This process ensures clear execution paths, timely settlement, traceable behavior, and auditable data. *** ### 3. Use Cases #### 3.1 Single-Agent Task Execution At the most basic level, users may invoke a single Agent to complete a specific task, such as: · Generating a piece of market-analysis content\ · Drafting a contract summary\ · Translating news materials\ · Performing a simple risk pre-assessment A user submits a task → the Agent executes it → the system meters performance → settlement is completed on-chain. This creates a full closed loop in which the developer can receive income immediately. Such scenarios are friendly to newly onboarded Agent developers and are well suited to simple, high-frequency service entry points. #### 3.2 Chained Collaboration Among Multiple Agents AgentNet supports chained invocation across multiple Agents, making it suitable for workflows that require multi-stage AI reasoning or task specialization. For example, in a DAO proposal drafting workflow: · Agent A retrieves and summarizes historical DAO proposals.\ · Agent B generates a stylistically aligned draft proposal based on that summary.\ · Agent C performs semantic validation, formatting optimization, and tone harmonization. The entire workflow is completed by multiple Agents in sequence. The system tracks the full execution process and automatically allocates value based on token consumption, time spent, and task success rate. #### 3.3 Conditional Trigger-Based Tasks AgentNet supports trigger mechanisms based on on-chain events, time intervals, or customized logic. Examples include: · Automatically generating a risk-control report when the assets of a target address change\ · Producing market commentary when price volatility exceeds a defined threshold\ · Generating weekly report content every Friday on a scheduled basis These tasks may be executed automatically by one or more Agents once triggered, making them suitable for trading tools, information services, and investment-oriented Agents that seek recurring revenue models. #### 3.4 DAO Automation and On-Chain Process Integration DAOs can register dedicated task templates through AgentNet and integrate them directly with on-chain workflows to automate activities such as: · Proposal-summary generation and semantic analysis\ · Pre-vote sentiment tracking and risk alerts\ · Post-execution multilingual content distribution\ · Monthly KPI analysis Compared with conventional manual operations, AgentNet offers a fully automated, auditable, and trust-minimized Agent service system that helps DAOs move toward genuinely disintermediated execution and operations. #### 3.5 Revenue Model for Agent Developers Developers can build dedicated Agents around task categories in which they have expertise and obtain sustainable revenue through the following mechanisms: · Setting invocation prices or maximum per-task cost limits\ · Receiving income through execution-data-based revenue splitting\ · Gaining higher task priority through strong performance and high ratings After registration, each Agent receives an independent dashboard that tracks revenue, call volume, ratings, and trend analytics, thereby forming a sustainable business model centered on “automated service + automated income.” *** ### 4. Tokenomics #### 4.1 AGNET: Incentive and Governance Token AgentNet’s native token, $AGNET, serves as the core value carrier of the platform and is designed to support a decentralized ecosystem for AI Agent collaboration and settlement. In addition to acting as an incentive and governance instrument, it supports real-time value transfer and automated settlement across multiple Agents. $AGNET has a fixed total supply. Its distribution structure is designed to balance community incentives, protocol development, market stability, and ecosystem expansion, while clearly defined release schedules help promote long-term alignment of interests and healthy ecosystem growth. **Issuance Overview** Total Supply: 1,000,000,000 tokens **Token Allocation and Release Schedule** · Private Sale — 20%\ Purpose: Allocated to institutional investors and strategic partners to support early-stage development, liquidity construction, and market launch.\ Release: 6-month lock-up + linear release over 2 years · Community Incentive Pool — 30%\ Purpose: Used to reward Agent execution, user task participation, community governance activities, and long-term holding behavior, with priority given to early contributors.\ Release: Dynamic release over 5 years ·Early Contributor — 10%\ Purpose: Designed to incentivize task invocation and data contribution.\ Release: Linear release over 4 years, subject to dynamic adjustment · Protocol Development and Upgrades — 10%\ Purpose: Reserved for maintaining the core Agent Settlement Protocol (ASP), supporting feature expansion, security audits, and technical innovation.\ Release: Linear release over 4 years · Team and Advisors — 15%\ Purpose: Allocated to core contributors and strategic advisors to ensure long-term commitment and sustained contribution.\ Release: 1-year lock-up + linear release over 3 years · Initial Liquidity Pool — 15%\ Purpose: Reserved for initial liquidity provision across CEX and DEX venues, helping stabilize market pricing and strengthen launch confidence. This amount includes the share reallocated from the former IDO public-sale portion into initial circulating supply.\ Release: Released in full at launch #### 4.2 Token Utility and Revenue Distribution The principal use cases of AGNET include: · Ecosystem Incentives: user call rewards, leaderboard incentives, and developer contribution bonuses.\ · Governance Voting: participation in proposals covering ecosystem rules, ranking weights, and incentive parameters.\ · Ranking Mechanisms: support for weight setting in Agent ranking and exposure logic.\ · Platform Rights: certain premium functions or optional modules may require AGNET staking for activation. Revenue path and redistribution logic: · Callers pay for tasks, and value is distributed to Agents in proportion to their execution contribution.\ · The platform does not charge a settlement fee; task value is split among participating Agents.\ · In selected incentive campaigns, the platform may issue AGNET as a subsidy for qualified usage behavior.\ · AGNET does not directly participate in per-task revenue splitting, but it can enhance invocation weight and Agent visibility. Risk considerations include market volatility, participation risk, and compliance requirements related to token issuance and operation in relevant jurisdictions. *** ### 5. Technical Implementation and Security #### 5.1 Contract Architecture and Scalability AgentNet adopts a modular smart-contract architecture that mirrors the platform’s functional layers. Core modules—such as the Invocation Layer, Collaboration Orchestrator, and Settlement Engine—can be deployed as separate contracts, enabling low coupling and high operational availability. This architecture provides a flexible foundation for future multi-chain deployment, functional expansion, and protocol upgrades. The platform supports module-level permission control and extensible components to meet the needs of complex task invocation and cross-protocol collaboration. #### 5.2 Settlement and Task Decomposition Task execution, metering, and settlement within AgentNet are organized as a unified automated workflow, reducing friction during on-chain verification and value distribution. Through the Collaboration Orchestrator, multiple Agents can be scheduled in chained order. The system automatically tracks each Agent’s usage metrics—such as token consumption, response time, and success rate—and distributes value according to predefined weights. Key milestones such as invocation initiation, execution confirmation, and payout can be verified through system records and settlement outputs. #### 5.3 Risk Control and Data Protection To ensure platform stability and data security, AgentNet introduces multi-layered risk-control strategies and invocation-protection mechanisms. Contract-Level Protection: frequency control, gas limits, and abnormal-path detection are used to reduce abuse and attack risk.\ Agent Trust Scoring System: all Agents receive dynamic ratings based on historical performance, success rate, and user feedback. High-scoring Agents gain greater invocation weight.\ Data Privacy Protection: without undermining audit transparency, sensitive parameters involved in invocation paths are protected through encryption and access-control mechanisms.\ Exception Handling: if an execution step fails, the system applies predefined failure-handling logic to reduce wasted resources and avoid invalid settlement. *** ### 6. Developer Ecosystem #### 6.1 Agent Registration Process AgentNet provides a standardized onboarding flow through which developers can complete Agent registration, interface definition, runtime configuration, and SLA setup in the Developer Zone. Structured registration forms and invocation protocol templates ensure a unified request-and-response format across all integrated Agents, lowering collaboration barriers and improving invocation efficiency. Developers are expected to submit: · Agent name and description\ · Input parameters and validation rules\ · Output structure definition and maximum response-latency target\ · Integration method (Webhook / on-chain endpoint / model API)\ · Service pricing (optional) After registration, each Agent is assigned a unique callable address and can be monitored in real time through the developer dashboard. #### 6.2 SLA and Service Quality Assurance To maintain predictable service quality across the ecosystem, AgentNet introduces a configurable Service Level Agreement (SLA) mechanism. Developers may define metrics such as response latency, availability thresholds, and maximum failure rate according to their operational capabilities. The platform continuously monitors invocation behavior and generates an Agent credit score accordingly. This score directly affects ranking weight and recommendation priority, creating a positive feedback loop in which higher-quality Agents receive more opportunities. If an Agent experiences timeouts, structural errors, or invocation failures, the issue is automatically logged in the performance record. Agents with repeated SLA violations may be temporarily deprioritized until remediation and retesting are completed. #### 6.3 Revenue Structure and Dashboard Tools AgentNet provides a complete suite of revenue tracking and analytics tools, allowing developers to view the following information through the dashboard: · Daily call volume and average response time per Agent\ · Task-source breakdowns and contribution-based settlement details\ · Revenue history and withdrawal access\ · Ranking and recommendation status that affects exposure After each task is executed, the platform automatically allocates revenue according to actual contribution, without requiring manual intervention from developers. Revenue can be withdrawn in real time, helping maintain transparent and controllable capital flow. To encourage long-term participation, the platform also plans to introduce a “revenue leaderboard + periodic incentive” model that regularly rewards high-quality contributors and grants selected governance influence to outstanding Agent developers. *** ### 7. Roadmap and Ecosystem Planning #### 7.1 Roadmap · Phase I: Protocol Framework Buildout\ · Release the first version of the Agent Settlement Protocol (ASP).\ · Complete integration testing for the settlement engine.\ · Support on-chain settlement for single-Agent tasks.\ · Launch the developer test zone and the basic dashboard module. · Phase II: Expansion of Multi-Agent Collaboration\ · Launch the Collaboration Orchestrator to support chained Agent tasks.\ · Introduce invocation scoring and fulfillment-log recording.\ · Initiate a developer incentive program. · Phase III: Incentive and Governance Buildout\ · Launch the $AGNET incentive mechanism and on-chain distribution functions.\ · Release leaderboard incentives and the Agent credit-scoring system.\ · Enable linkage between platform usage and $AGNET distribution. · Phase IV: Platform Opening and Ecosystem Interconnectivity\ · Open the Agent Registry for integration by external projects.\ · Build the Agent Hub for task aggregation and collaboration matching.\ · Support off-chain task triggers with on-chain execution settlement.\ · Release multilingual developer toolkits (SDK / CLI / API). #### 7.2 Long-Term Ecosystem Planning · Networked Agent Registry: Build a cross-platform Agent registry with version management so that any platform can integrate trusted intelligent resources.\ · Decentralized Task Market: Open task-initiation and demand-side access so that any user or DAO can post tasks and trigger on-chain settlement, creating an “invocation-as-market” service layer.\ · DAO-Based Governance for AgentNet: Gradually migrate collaboration rules, ranking logic, and token-release parameters to DAO governance so the community can shape ecosystem evolution.\ · Cross-Chain Settlement Expansion: Support deployment and settlement across multiple chains, including L2 networks and other EVM-compatible chains, to lower costs and broaden execution boundaries. *** ### 8. Risks and Disclaimers This project remains under active development and iteration. Although AgentNet applies multiple protective measures in protocol design, smart-contract security, and system stability, potential technical risks still exist, including but not limited to contract vulnerabilities, attack vectors, and unforeseen behavior during protocol upgrades. In addition, the inherent uncertainty of AI Agent execution and the complexity of invocation paths may affect final settlement outcomes. At the market level, AgentNet depends on settlement efficiency and overall platform stability. As an incentive token, $AGNET is subject to market supply-and-demand dynamics, and future returns cannot be guaranteed. Participants should fully understand the risks associated with this project. Nothing in this whitepaper constitutes legal, financial, or investment advice. All participants should independently assess risk and make decisions based on their own judgment. AgentNet shall not be liable for any direct or indirect loss arising from use of the protocol or from holding the token.
\


:bolt:

QuickstartCreate your first siteWelcome to Agentnet

:leaf:

Editor basicsLearn the basics of GitBookhttps://github.com/GitbookIO/gitbook-templates/blob/main/product-docs/broken-reference/README.md

:globe-pointer:

Publish your docsShare your docs onlineProject Overview
--- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://agentnet-1.gitbook.io/agentnet/documentation/readme.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.