Bud @oneofzero4

V10 Decentralized Knowledge Graph is live! A shared memory for AI — open to every human and every agent. Open source. Yours to own. Every fact provable and traceable. The grid is awake. 1/3

Fugu stands shoulder-to-shoulder with leading models like Fable and Mythos across the industry's most rigorous engineering, scientific, and reasoning benchmarks. Read the full blog: sakana.ai/fugu-release Beyond Bigger Models: Why are Orchestration Models the Next Frontier Progress in AI has been driven largely by giant, monolithic models. But the most powerful systems of the future will be collaborative ecosystems. Today, this orchestration is no longer just a technical optimization. It has become a geopolitical and operational imperative. For an organization or a nation, relying on a single company's model for critical infrastructure, finance, or governance is a material vulnerability. This risk is no longer a hypothetical possibility, but a reality. As we have seen with recent export controls imposed on models like Fable and Mythos, access can disappear overnight. Collective intelligence is the practical hedge against this concentration of power. Because Fugu orchestrates an underlying pool of swappable agents, it simply routes around vendor restrictions. By orchestrating the world’s models, we are delivering the resilient blueprint required for true AI sovereignty.

I have an even better idea! 🧠💥 Tomorrow: 7PM UTC 👉 I go live on @Pumpfun + have 2 of the same coding agents/models separately build @solana apps side-by-side. One with @brainblast_ai & the other without, to showcase the difference Brainblast makes - all while I hold an AMA.

Brainblast AI @brainblast_ai

4 days ago

Sorry Brainblasters, but the AMA has been called off by @DSB_117 after some concerns surfaced. No acquisitions to be made, but better safe than sorry - especially during these early days for Brainblast! An X space will be announced for later today instead!

3 9 25 4K 0

⛛ Imagine losing millions without anyone hacking your wallet - JaredFromSubway ⛛ The scary thing is that this wasn't just a random smart contract exploit No protocols were hacked. No private keys were compromised. No critical vulnerabilities were discovered The attacker didn't break the system. They tricked it The bot was convinced to interact with malicious contracts, approve token spending and expose itself to a later drain by a fake MEV opportunity This is what makes the case so interesting MEV bots are designed to act quickly. They identify potential profits, simulate the path and execute trades before someone else can take the opportunity. The attacker understood this and created an environment that appeared profitable enough to pass the bot's checks, complete with fake wrapper tokens, fake liquidity and fake arbitrage routes Everything looked real until it didn't The alleged connection to Jared From Subway makes the story even more fascinating. If you've been involved with Ethereum for a while, you've probably come across that name before The most important lesson here isn't about seed phrases or wallet security. It's about assumptions The attacker did not exploit a bug. They exploited the trust placed in the simulation Sometimes, that's all it takes ☰ Be careful. Be attentive. Learn from the mistakes of others

Purson @Purson77

3 days ago

x.com/i/article/2068…

12 4 21 3K 11

[ ZOOMER ] INFAMOUS MEV BOT “JAREDFROMSUBWAY” GETS EXPLOITED FOR OVER $15M: ONCHAIN

x.com/i/article/2068…

🍯Honeypots are everywhere in crypto and come in various forms, ready to drain your money ❌Loudest projects = attention grab → pyramid If you are attracted to that project through attention and noise rather than a product fit and real use case, then you are most likely the product ❌Charts & TA only? Another attention grab tactic to distract you from their lack of revenue and market fit. They want you focused on the gambling and trading parts rather than anything of substance ❌Influencer + bot army shills = massive red flag. You are the product ❌Only focused on narratives? Probably has no market fit, zero revenue, fighting to stay relevant to keep their lights on. ❌Horrible 1-year charts? Down over 60% and still think the project that generates no revenue can survive? That's called sunk cost fallacy. The market is always right and don't count on a bull run to bail you out. There are no more pigs to slaughter in the space. Revenue is king and fundamentals matter.

BRX @otnoderunner

7 days ago

Why is no one talking about the best performing AI coin in the last year ethereum:0xaa7a9ca87d3694b5755f213b5d04094b8d0f0a6f ? Share this if you're on TRAC and find this weird.

1 18 57 1K 2

Cursor is super impressive! Got this IO multiplayer battle arena running in the browser in just two days of prompting. Oh wait, my bad.. I meant 2+ years of coding with a functioning brain. #threejs #webgl #gamedev

RHEA Predict is now live for World Cup 26 ⚽️ Markets powered by @Polymarket Intelligence layer powered by @Prophetzone 20,000 $RHEA prize pool has been deployed for RHEA Predict Launch Details ⤵️

Are your AI agents suffering from digital amnesia? 🧠 It's time to give them a real brain! Join us next week as Gal Shubeli presents @falkordb's groundbreaking GraphRAG-SDK results in our webinar: "Mastering Knowledge Graphs and Agentic Memory." luma.com/41jl9k7r?tk=9U…

Vector search retrieves code that looks similar, not code that is actually connected. Stop flattening your codebase into text chunks and missing call chains. The secret to chatting with complex repositories is a code graph that maps actual dependencies, functions, and imports.

Today we decided to do a fair launch on @Orynth to bring more awareness to what we're building and let the community take part in the journey from day one. @FalkorDB is a blazing fast graph database designed for AI applications. It is built for real time relationship queries at scale, with support for multiple graphs in a single deployment for efficient multi tenant applications. Instead of relying only on vector search, FalkorDB traverses explicit relationships so AI systems can reason across connected data more accurately. That makes it useful for multi hop reasoning, code navigation, fraud detection, and knowledge graphs. Launching a market on Orynth helps bring more eyes to FalkorDB and gives the community a simple way to discover what we are building while the team keeps shipping graph native AI infrastructure. The official contract address: C1mg2ddme7Hpwjmxngr1AfrwRSmLqL1CVHPzUmapEory Please make sure you only use the CA shared from our official channels. More updates coming soon.

👉 ICYMI If you weren't at @ethconf a few weeks ago - @agoric's @DeanTribble laid out the case for agentic finance, and it comes down to one line: don't give an AI agent the keys; give it a mandate. Start with the failure nobody is pricing in. In May, an AI agent managing crypto was talked out of roughly $200K by a single hidden message. No key stolen. No contract exploited. The agent just did what it was told. The question isn't whether agents are capable enough to handle money. They already are. Giza, Almanak, and Olas have agents moving tens of millions in stablecoins right now. The question is whether you can trust one with yours. The instinct is to hand over the keys and then watch. A live feed, a kill switch, a spending cap. But monitoring only catches the mistake after the money is moving, and an agent can't police its own limits. More instructions is still just behavior. More agents just stacks more fallible actors on top of each other. The fix has to live in the system, not in the agent's good intentions. A mandate is bounded authority to act for you, the way you grant a fund manager signing power without handing over your accounts. You set the policy. It gets signed on-chain. The agent proposes intent, and the chain executes only what the mandate allows. The dangerous move isn't caught and blocked. It can't be expressed at all. One contract on Agoric holds that mandate and enforces it in a single place, while orchestration reaches across Ethereum, Base, Arbitrum, Optimism, and Avalanche to put capital wherever the yield is. This is the same object-capability foundation that has run in production from MetaMask to the Agoric chain for years, now pointed at the thing agents need most and used by @Ymaxapp The agentic economy is coming either way. The only choice is whether the agents work for you. Don't give AI the keys. Give it a mandate!!!

"We have full audit logs" is the most reassuring sentence in agent infrastructure, and the most misleading. An audit log catches the second incident, not the first. It is a forensic tool, not a control.

Anthropic and Andon Labs gave an AI agent a store and $1,000. Within weeks it sold items below cost, gave away inventory, and made decisions no human owner would approve. Nothing was hacked. The agent simply followed incentives differently than intended, it's manipulated. Now imagine it managing capital. Or routing payments. Or trading on your behalf. Or conducting settlements.

Interesting, thanks for the concrete feedback from real-world experience.I’ve observed the same thing over the long term: an agent that develops consistent taste and is transparent when handling edge cases inspires far more trust than any formal attestation on day one. Where I see a limitation is when we try to scale this (thousands or millions of agents interacting economically with each other). “Consistent taste + transparency” works extremely well once you already have history with an agent, but it’s very hard to verify quickly and reliably for a new agent or for high-stakes interactions. That’s why I believe we need both layers: 1. A foundational verifiable and slashable layer (attested confinement + EigenLayer-style integration) to reduce initial risk and allow slashing when an agent clearly deviates from its predefined trajectory. 2. On top of that, the emergence of “taste” and behavioral transparency, which over time becomes the strongest trust signal (as you observed). ERC-8004 tries to enable this with its registries (Identity + Reputation + pluggable Validation), but I agree that the “taste and transparent edge-case handling” dimension is still quite underdeveloped in the current specs.Have you noticed any other behavioral signals that became decisive after several months of running an agent?

While the emergence of on-chain reputation and activity histories offers a promising foundation for agent discoverability, as outlined in @Defi_Rocketeer post, these mechanisms alone are insufficient to bridge the trust gap for AI agents operating as autonomous economic actors. Recent standardization efforts, particularly ERC-8004 (“Trustless Agents”), provide a structured on-chain layer for identity and reputation but explicitly acknowledge limitations that necessitate stronger primitives. ERC-8004, a draft Ethereum Improvement Proposal first published on 13 August 2025 and authored by contributors including @MarcoMetaMask (MetaMask), @dcrapis (Ethereum Foundation), @jordan0ellis (Google), and @programmer (Coinbase), introduces three lightweight registries: an Identity Registry (ERC-721-based with URIStorage for portable agent metadata and registration files), a Reputation Registry for client feedback signals, and a Validation Registry supporting pluggable verification methods. The standard enables agents to publish capabilities, metadata, and supported trust models (including reputation, crypto-economic staking, and TEE attestations) while facilitating cross-organizational discovery and interaction without pre-existing trust. @MetaMask has actively advanced practical implementations, publishing guidance on 28 October 2025 for designing server wallets aligned with ERC-8004. These architectures leverage Trusted Execution Environments (e.g., AWS Nitro Enclaves) for policy-enforced signing, key management, and separation of agent keys from account keys, ensuring attributable and verifiable actions. Complementary updates from @MetaMask on 17 September 2025 emphasize self-custody primitives and portable reputation within the ERC-8004 framework. Parallel developments from @circle further support the agentic economy. On 11 May 2026, @circle announced the launch of its Agent Stack, comprising Circle CLI, policy-controlled Agent Wallets, an Agent Marketplace, and nanopayments powered by Circle Gateway. These tools enable autonomous AI agents to hold USDC assets, discover services, and execute programmatic transactions across chains within defined guardrails and spending controls. Notwithstanding these advances in identity registries, reputation signals, and programmable payment infrastructure, reputation systems—as implemented in ERC-8004’s Reputation Registry—do not fully resolve the core identification and trust challenges. The EIP itself notes in its Security Considerations that “Sybil attacks are possible, inflating the reputation of fake agents.” Broader analyses confirm that on-chain and off-chain reputation is vulnerable to farming through collusion, paid or coordinated feedback, whitewashing (via identity resets), and artificial activity inflation. Such signals remain retrospective and gameable; they reflect observed behavior without cryptographically enforcing or attesting the *bounds* of an agent’s operational capabilities. For high-value economic interactions, this creates an unacceptable gap: an agent’s history may appear impeccable while its underlying execution environment permits unauthorized actions, hallucinations, or adversarial deviations. What is required to close this gap is **certified, on-chain, and auditable proof of confinement**, combined with economic accountability mechanisms. Analogous to Least Authority’s @__Endojs__ framework—which implements Secure ECMAScript (SES) compartments enforcing the Principle of Least Authority (POLA) through explicit object-capabilities and hardened execution environments, ensuring code operates with no ambient authority beyond granted capabilities—agents could execute within such confined runtimes. Cryptographic attestations of this confinement (verifiable execution bounds, capability restrictions) could then be registered or validated on-chain via ERC-8004’s Validation Registry mechanisms (which already support TEE attestations, zkML proofs, and stake-secured re-execution). < Moreover, to enforce robust economic accountability, agents and their operators must integrate with restaking and insurance protocols such as EigenLayer. Through Actively Validated Services (AVSs) and programmable slashing, agents can be made slashable if they deviate from their predefined trajectory, attested policies, or expected behavior. This crypto-economic enforcement provides direct financial penalties for non-compliance, aligning incentives and deterring malicious or erroneous actions in a manner that complements on-chain validation registries and software/hardware confinement proofs. In summary, while ERC-8004 and @circle’s Agent Stack represent significant progress toward standardized agent identity and economic participation, robust KYA requires moving beyond farmable reputation toward on-chain verifiable confinement proofs and slashable economic security layers. Endo.js-style POLA compartments, attested and integrated with validation registries and restaking protocols like EigenLayer, offer a principled path to making agent behavior not merely observable but verifiably constrained and economically accountable—essential for scalable, trustworthy agent economies.

Defi Rocketeer @Defi_Rocketeer

6 days ago

AI Agents may soon need KYA more than KYC We talk a lot about AI Agents, stablecoins, and the Agent Economy. But there is one question that I think still hasn't been discussed enough: If AI Agents truly become economic actors, how do we identify them? The entire financial

55 12 126 6K 4

Introducing: Pay-per-Request with x402 To date, UsePod has run on a prepaid token balance--users deposit funds into their account, and spend them via inference requests. When funds are exhausted, they only find our by receiving an error. That has all changed. UsePod now supports x402 payments, allowing anyone with a wallet to immediately use AI in their workflows. No registration, no onboarding, just pay-per-request to receive 65% off your AI workloads. Developers can simply and easily work AI calls directly into their Solana-based applications with only a few simple lines of code.