.@GenLayer last week: @courtofinternet × @MetaMask: an AI jury can revoke an agent's Smart Account delegation if it acts against the user's intent Tested on 179,842 past @Polymarket markets, every resolution matched the real outcome Clarke testnet is the release candidate. Mainnet + GEN TGE targeted for Q4 2026 Justice needs an upgrade

HOW IT IS WORK? @lumera in 30 seconds: validators on top, SuperNodes underneath, proofs in between. Consensus above. Work below.

.@monad, the week (Oct 12–18) DeFi TVL is holding near $1B at ~$978M. monad:native slid from $0.027 to $0.022, and the work didn't slow down. Private Settlement explained: private execution domains on the L1, settling against shared liquidity nONYC lands via Plume Vaults, bringing OnRe's reinsurance backed yield to EVM Category Labs publishes Tree-BTE, groundwork for post-quantum encrypted mempools CertiK snapshot: ~$1B TVL, $2.64B in 30d DEX volume, $1.11B in lending supply Metropolis submissions closed. Blitz ran in Medellin and Shanghai

Celestia, mid October 2026: the build is real, and usage is catching up celestia:native is up ~59% in four weeks, its highest level of 2026, and still ~97% below its 2024 high. Here is what drove it Derive V3 went live on Oct 8. A top onchain options venue, with $14.2B in notional this year, settles on Ethereum and posts its state data to @celestia Fibre benchmark: 3.07 Tb/s across 120 validators, with peaks of 4.27 Tb/s. Mainnet starts at up to 128 MiB, and activation still needs validators to signal Tokenomics: issuance is ~2.5% and falling toward 1.5%. The Sustainable Blob Economy proposal aims to replace part of it with paid demand. Adoption is real. Usage has to follow

BlobPedia, our @celestia knowledge base for AI agents, just got three new answers Which server to rent. 1,427 servers from 23 providers, each with a verdict for every node role (validator, bridge, Fibre) and the reason. The verdicts are Pulse's own, so your agent never guesses Whether the provider has trouble. Every server detail lists the provider's open status page incidents Whether Fibre is ready. On Mocha on 13 Oct, 82 of 85 validators were ready, holding 99.73% of stake against the 66.67% needed. Losing 6 validators, 3 providers or 3 countries would break 2/3. Data from @Cumulo_pro It all runs through one MCP endpoint, and every new answer names its source. No key needed. claude mcp add --transport http --scope user blobpedia blobpedia.org/mcp It also works in Claude.ai, Cursor, Windsurf and VS Code.

Monad, the week (Oct 5–11) @monad Private Settlement: Category Labs published a design for confidential execution domains on the public L1. They are enabled by a standard EVM contract, with encrypted transactions, state and code, and atomic settlement against shared Monad liquidity. Built for banks, funds and treasuries. Cadence demoed at Open Singapore: multiple concurrent proposers, targeting ~100 ms blocks and ~300 ms finality Typewriter is out, open source: custom sequencing, millisecond confirmations, gas sponsorship, built in accounts $MON is now on @RobinhoodApp. @plumenetwork Nest Vaults have landed on Monad Privacy for institutions, speed for everyone

Just submitted Kudy to CLOCK IN, a Solana Mobile hackathon presented by Radiants. kudy.unitynodes.com/ @solanamobile

Content moderation you can audit. Modeq: GenLayer validators classify each post with an LLM, a majority has to agree, and fixed thresholds in the contract decide ALLOW, FLAG or BLOCK. The model never gets the final call. modeq.unitynodes.com Built on @GenLayer Intelligent Contracts.

Celestia, the week. Fibre, @Celestia's high-throughput data availability protocol, posted its end-to-end benchmark: 3.07 Tb/s on average across 120 validators, verified onchain. Peaks hit 3.69 Tb/s over 60 seconds and 4.27 Tb/s over 30. The road to mainnet: app v10 is live on Mocha and validators are setting up Fibre servers. Mainnet starts at up to 128 MiB and scales toward 3Tb/s and beyond. No date yet. Derive V3, the onchain options protocol with $14B+ notional in 2026, is moving its data availability to Celestia. Proven at 3 Tb/s, heading to mainnet.

Lumera, three weeks in @FalconXGlobal, 1XP and NTT DOCOMO GLOBAL joined the validator set, and the team says more are coming A Lagrange partnership: zkML proofs for AI inference are now kept permanently in Cascade. And on 29 September, Seoul Builder Night before Korea Blockchain Week: the Intelligence Layer, SuperNodes, EVM for builders and @cascade_xyz as memory for agents. Memory for agents, backed by institutions