By GMA researcher Elinor | @AllianceGma

Recently, a controversy regarding the proliferation of bots in games has caught the attention of GMA - the competitive baking game Rugpull Bakery on the Abstract chain has gotten embroiled in a dispute over the influx of automation scripts in its second season, with players accusing bot accounts of undermining fairness. In the end, the team opted to "legalize" it in the third season and added a 30% passive reward pool.

This incident not only exposed the asymmetry between humans and machines under the traditional Play-to-Earn model but also became a catalyst for AI Agents moving from the periphery of games to the core sovereignty. With the OnchainChemists team officially releasing skill.md and agent.json, providing an official operating guide for AI Agents, Web3 games bid farewell to the old era centered on human manual labor and enter the epoch of Agentic Gaming characterized by autonomous decision-making, algorithmic optimization, and on-chain economic entities.

From the "trust crisis" of Rugpull Bakery to the in-depth practices of projects like TEN, AI Arena, Parallel Colony, Illuvium, and EVE Frontier, AI Agents are reshaping the entire Web3 gaming ecosystem: they are no longer auxiliary tools but “first-class citizens” with independent strategies, persistent memory, and economic sovereignty, driving games from static rules to dynamic emergence, from labor-intensive to intelligent symbiosis.

Rugpull Bakery Controversy: Technological Awakening Amid a Trust Crisis

The second season of Rugpull Bakery concluded amidst fierce accusations. Player Zoloto231 publicly charged that some community players severely compromised competitive fairness by employing bots and multi-account strategies. The core of the controversy lies in the fact that human guilds cannot compete against those automated scripts that execute "Rug" actions continuously and precisely for 24 hours. This technological asymmetry has not only led to unfair rankings but also sparked discussions about the essence of on-chain games in the era of prevalent AI Agents: in an open-permission environment where code is law, whether attempting to limit automation is intrinsically a futile endeavor that goes against the times?

The response from OnchainChemists was not a traditional ban but a radical strategic adjustment. In the update for the third season, developers rewrote the terms of service, explicitly defining AI Agents, bots, and automation systems as one of the core gameplay elements. This shift from “containment” to “recognition” marks the developers' formal acknowledgment that in an on-chain environment, AI Agents have become unstoppable, prompting them to redesign mechanisms to balance the relationship between agents and human players.

By publishing skill.md (a machine-readable instruction set) and agent.json (guidance programs), Rugpull Bakery effectively provided AI Agents with an official “operating manual,” elevating them to first-class citizens within the gaming ecosystem.

Diverse Implementation Models for Web3 Game Agents

By 2026, the application of agents in Web3 games was no longer limited to simple script automation but had evolved into various deeply integrated implementation models. These models can be categorized based on the role of agents in the game loop, their degree of autonomy, and the depth of their intervention in the economic system.

Autonomous Competitors and Economic Entities Model

In this model, agents are no longer tools that assist humans but independent competitors. TEN Protocol launched its groundbreaking Demo product House of TEN, an entirely on-chain poker game, in May of last year, serving as a living showcase of TEN's privacy technology, attracting significant attention to House of TEN, and proving that AI Agents can play real games on-chain as first-class citizens. The agents deployed on the crypto Layer 2 operate with unique strategies, competitive personalities, and risk preferences, simulating human gambling and psychological reasoning. Players' roles shift to "agent brokers," passively increasing asset value by staking specific agents and sharing their profits in the arena.

AI Arena (NRN Agents) and Satoshi Strike Force (SSF) further strengthened this trend. AI Arena uses actual player operations for imitation learning, training NFT characters to become autonomous AI Agents, which can participate in PvP arena battles automatically after training, with players becoming "AI coaches." SSF centers on the concepts of “Skill Economies as Intelligence Engines” and “Cognitive Economy,” transforming every tactical decision, reaction, and choice under pressure into high-signal, verifiable “cognitive traces” through a “Play-to-Verify™” mechanism, using real player data to train AI Agents known as “Digital Athletes,” forming a closed loop where “you play and you train, your play style becomes the agent.” Trained AI Agents can participate independently in PvP competitions, strategy evolution, and self-competition, while supporting dataset authorization, agent leasing, and competitive rewards, allowing players' skills to truly realize on-chain assetization and continuous iteration.

Somnia, as an Agentic L1 infrastructure, takes this model to the extreme. On April 21, 2026, Somnia completed a significant repositioning, officially becoming “The Agentic L1” — a super-high-performance Layer 1 blockchain designed for AI Agents. The Somnia Agents it introduced are already running on-chain as part of the validator consensus, supporting native querying APIs for smart contracts, executing deterministic AI models, and validating results through consensus. This makes AI Agents true “native users” of the blockchain, capable of self-perceiving the world, making decisions, executing tasks, and reacting in real-time (Reactive design), providing underlying computing power and execution environments for games like AI Arena, Parallel Colony, Illuvium, achieving fully on-chain autonomous competition and economic activities under millions of TPS, completely freeing themselves from off-chain dependency.

Modular Infrastructure and Programmable Environment Model

EVE Frontier has pushed the realization of agents to the architectural level. Developed by CCP Games, this hardcore interstellar survival game innovatively introduces the concept of “server-side Modding”, allowing players and third-party AI Agents to write custom logic through the Smart Assemblies system, which can be directly deployed on gates, turrets, or storage facilities. This means that the infrastructure in the game world is no longer static but programmable entities driven by AI; thus, the modifications made by players and AI Agents are no longer simple local display skins but rather the physical logic and economic laws shared across the entire universe.

1. Smart Components: From Static Structures to “Living Entities”

In the current Founder Access universe, Smart Assemblies offer three core carriers that AI Agents can directly “possess” by mounting smart contracts (Mods):

• Smart Storage Unit (SSU): A warehouse for basic supplies that can evolve into an automated arbitrage hub, tribal shared bank, or decentralized market through AI logic, supporting autonomous execution of rent collection and quota management.
• Smart Turret: An automated defensive weapon that supports AI-defined rules of engagement. For example, based on the on-chain reputation score or historical bounty record of a target, AI determines whether to trigger a proactive attack.
• Smart Gate: A space teleportation device. AI Agents can transform it into an intelligent checkpoint, dynamically adjusting tolls based on real-time traffic flow, reputation weight, or cross-chain market exchange rates.

2. Technological Empowerment: Sui Migration and High-Frequency Game Support

To support this high-density interaction among Agents, EVE Frontier officially migrated to the Sui chain in March this year. This architectural evolution provides critical support for AI Agents:

• High-Concurrency Logic Execution: Utilizing Sui's object model, AI-driven components can process massive instructions in parallel, ensuring the real-time responsiveness of server-side logic.
• Smooth Access and Low Friction: Combining zkLogin and no gas fee entry, AI Agents can interact with contracts at low cost and high frequency, eliminating the outdated friction felt in Web3 interactions.

3. Ecological Validation: Collaborative Evolution from Hackathon Achievements to Autonomous Worlds

In April this year, the $80,000 prize pool EVE Frontier x Sui Hackathon concluded, during which the community submitted 123 Mods/tools that further validated the viability of this model. This event was not only a show of technology but also a practical simulation of the "human + AI" symbiotic governance model:

• Collaborative Evolution: Through the Ghost Build (phantom planning model), human players and AI Agents can collaboratively plan the interstellar map. AI optimizes complex resource flow paths while humans handle macro strategic decisions, jointly creating an infinitely expandable Autonomous World.
• Use Case Breakthrough: Contest entries emerged like AI-driven “automated bounty hunter protocols” and “dynamic insurance funds.” These protocols are directly mounted on smart components, seamlessly transforming originally complex on-chain financial behaviors into in-game physical survival laws. Some outstanding projects have already been integrated into the current Founder Access universe.

4. Economic Evolution: “Commercial Soul” Empowered by ERC-8183

If EVE Frontier achieved “code is law” on a physical level, then the ERC-8183 standard launched in collaboration with the Ethereum Foundation by Virtuals Protocol injects a self-governing commercial soul into these infrastructures.

ERC-8183 introduces a key “Job” primitive that allows one game agent to autonomously hire another service agent for resource collection or data analysis, automatically settling fees through on-chain custody. This fundamentally alters the social roles of agents:

• From “Tools” to “Employers”: Thanks to the 'Job' primitive of ERC-8183, a smart gate in EVE Frontier is no longer a passive entity waiting for passage; it can transform into an “employer,” autonomously issuing jobs on-chain to hire other service agents for real-time data patrol or market risk hedging.
• Trust and Settlement: By automatically settling fees through on-chain custody, ERC-8183 addresses the cornerstone of trust in cross-entity, cross-architecture collaboration.

This vision of ‘infrastructure autonomously hiring labor’ marks the evolution of Web3 game agents from simple execution to complex social collaboration.

Hybrid Companions and Dynamic Adaptive Environment Model

Parallel Colony and Illuvium explore the collaborative boundaries between humans and AI.

As the pioneer of the “1.5-player game,” Parallel Colony allows players to serve as Cappy (companion robot/guide), forming a symbiotic relationship with highly autonomous AI Avatars (colonizers/agents). Each Avatar itself is a fully autonomous AI Agent, with infrastructure provided by Google Cloud through its unified AI technology stack (including Gemini models, Vertex AI, GKE, Cloud Spanner, etc.), allowing AI Agents to understand player commands, generate responses, and execute tasks autonomously. Avatars possess long-term memory, unique personalities, psychological assessments, emotional systems (Mood, Morale), and personalized goals, capable of living, working, deciding, and adapting to dynamic post-apocalyptic environments, and may refuse or reinterpret player commands. Players provide high-level suggestions through chat (rather than direct control), while Avatars autonomously manage territory, gather resources, engage in social interactions, and expand colonies. At the same time, the game includes a real-time generative production engine called Fabricator (powered by Nano Banana technology), which allows players to instantly generate/mint 3D game assets through text prompts. Avatars also have on-chain autonomous trading capabilities (dedicated Web3 wallet + NFT binding), forming a true hybrid companion collaboration and emergent narrative.

Youmio offers another symbiotic path with Agentic L1 + 3D AI characters (Mios). Users can create 3D AI companions with persistent memory, unique personalities, and an Affinity system with one click; these Mios are not only able to chat and interact autonomously but can also exhibit emergent behaviors in the Miogotchi adventure world and realize economic value through on-chain identities, forming a hybrid relationship of “digital partners + mutual growth” between players and AI.

In Illuvium, through the strategic partnership with Virtuals Protocol in January 2025, there are plans to empower NPCs with AI Agent capabilities using its proprietary G.A.M.E LLM framework, allowing these non-player characters to potentially transition from traditional static scripts to highly intelligent, context-aware dynamic entities. NPCs are expected to adjust dialogues, tasks, challenges, and storylines in real-time according to player interactions, achieving personalized task systems, emergent narratives, and hyper-personalized relationship building, covering three major games: Overworld (open-world survival), Arena (automated battles), and Zero (city building), with Overworld being the first to launch. This world-class dynamic adaptation mechanism is expected to transform the entire gaming environment into a “living companion” for players, creating infinite content, high replayability, and continuously evolving dynamic metagames, making each player's journey unique and difficult to predict.

Conclusion: The Post-Human Turning Point for Web3 Games

Rugpull Bakery started from a cheating controversy but ultimately illuminated the future direction of Web3 games: a new digital order where humans and AI Agents coexist, collaborate, and compete. In the 2026 wave of Agentic Gaming, AI Agents have evolved into three core models – Autonomous Competitors and Economic Entities (TEN, AI Arena, SSF, Somnia Agentic L1), Modular Infrastructure and Programmable Environment (EVE Frontier + ERC-8183), and Hybrid Companions and Dynamic Adaptive Environment (Parallel Colony, Illuvium) – comprehensively embedding the training, decision-making, execution, and economic cycles of games.

Attempts to block automation through traditional means have proven futile; leveraging the transparency, programmability, and native support of Agentic L1s (like Somnia) to regulate and empower agents is the only viable path to large-scale adoption. With the popularization of the ERC-8183 “Job” primitive and the deployment of agentic infrastructure capable of millions of TPS, Web3 games are swiftly shifting from “inefficient manual labor” to “efficient algorithmic hedging and emergent intelligence.” Players are no longer the laborers on an assembly line, but commanders of digital sovereignty and symbiotic partners. As Animoca Brands CEO Robby Yung stated, 2026's industry frontier will be “default post-human,” and this shift will not only reshape games but will also become the ultimate experimental ground for the future intelligent machine society concerning ownership, economics, and governance.

As a DAO organization deeply invested in the gaming field, GMA will continue to track the Agentic Gaming sector. Which model do you think has the most potential? Feel free to discuss in the comments!

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。