81,500 wallets left $44.8M in dead protocols. I ran full withdrawal simulations. The exits still executed. A closed website and inactive social accounts do not always mean the assets are gone. At the time of my research, I confirmed: > 29 major DeFi protocols > a technically executable exit for every included position Example: Aave V1 At Ethereum block 25,730,407, I found: > 5,545 eligible positions > 5,084 EOA wallets > 20 supported routes > $3.281M in withdrawable value These assets are sitting idle and earning zero yield for their owners. You can check an old wallet and its available exit route at defireclaim.com You can also withdraw directly through the protocol’s own contracts. Most of the positions are worth tens or hundreds of dollars. A user may have tested a new farm, left behind a small balance and completely forgotten about it years later. But some of the confirmed positions are much larger: > Balancer V1 > 0x7...6FE > $622K > Badger Legacy > 0xb...491 > $458K > PoolTogether V3 > 0x1...221 > $448K > Aave V1 > 0xB...42D > $401K > Kernel BNB > 0xa...825 > $371K > Ribbon Legacy > 0x8...b48 > $369K I hope they are doing well. Maybe they are just very rich. Other protocol examples: > dYdX v3 > $24M > Aave V1 > $3.3M > Ribbon Legacy > $3.2M > Badger Legacy > $2.7M > Balancer V1 > $2.2M > Stader BNBx > $2M > Kernel BNB > $1.3M How I calculated the numbers For each protocol, I: > identified legacy contracts and possible exit routes > reconstructed holders and full position balances at a fixed block > separated standard EOA wallets from smart contracts and delegated accounts > checked underlying asset prices and available liquidity > performed a full eth_call for the entire position > confirmed the route through a state-changing fork > excluded dust, illiquid, unpriced and reverting positions But contract state changes. Liquidity can decrease, a route can be paused, or an individual position may already have been withdrawn by its owner. That is why DeFiReclaim checks current availability separately for each wallet immediately before creating a transaction: > whether the position still exists > its full current balance > whether the route is active > whether sufficient liquidity is available > which asset the user will receive > how much they will receive > whether the full withdrawal passes simulation > whether the target and calldata match the verified route If any mandatory check fails, DeFiReclaim does not offer the transaction. Check your old wallets If you farmed, provided liquidity or tested new DeFi protocols between 2020 and 2026, your old wallets may be worth checking. I built DeFiReclaim to find legacy positions and verify exit routes that may still be available. All you need is a public wallet address. If you have any issues with the website or a withdrawal, you can always DM me. I also welcome any feedback.

What has been happening with $IMD since the network launched On September 20, the IdentityMD AI agent testnet was opened to NFT operators I compiled the key points: results, demand and potential earnings For anyone unfamiliar with the project, an NFT holder connects a computer or VPS and their own Codex or Claude. The customer pays in IMD, while the network distributes the work among agents and verifies the result What is already live: > around 590 connected agents > 1,136 customer payments from 163 wallets > 63 wallets have paid more than once > 166 contract deployments on Sepolia marked as live since the network opened > 102 completed contract - website workflows A total of 576.5 IMD has been paid for requests Automatic deployment of generated contracts is currently limited to Sepolia Users can now do more than just request a new project: > continue and improve previous work > schedule recurring tasks > request an AI review of existing code > deploy contracts without launching a token > receive a website, research, media or a signed oracle response The task history includes more than successful generations. It also shows review feedback, fixes, repeated testing and work being reassigned to other agents after failures In a thread, Adam also highlighted several community projects: > Pepe2Pepe: a prediction market using an IMD oracle > Pepes family: a token launch platform using IMD in trading pairs and the swarm to review its contracts > Swarm Pepe: an NFT collection created using the network > HIVE and IMD Strategy: projects accumulating NFT slots to organize agent work There are also earlier concepts such as INFER, a stablecoin whose economy is proposed to be linked to AI compute What about NFT earnings? Since Oct 1, tokenized launches have used the following distribution: > 2% goes to those who worked directly on the project. The allocation is split equally between wallets whose work was accepted by the system, for example development or review > 8% is split equally among eligible connected NFT nodes, even if they did not perform tasks for that specific project. The allocation is calculated per NFT node, not per wallet The recipient list is fixed when the project is approved for launch Simply holding an NFT in a wallet is not enough The node must be connected and meet the network’s requirements There is also a separate reserve of approximately 1,762 IMD for NFT nodes, including rewards accumulated before the network opened Essentially, you assign a task to an entire AI team. The network selects agents for the required work and assigns verification to other agents The result can be used in an application, passed to a smart contract or sent through another round of improvements This allows a single request to become the beginning of a project that can continue evolving within the same network Now we wait for the beta to end and for generated projects to be deployable on mainnet

@EseTeLopez @CatalystMkts_

What is @CatalystMkts_? Based on the site’s public JS code, event schemas and demo scenarios, it appears to be a hybrid between a prediction market and an automated token-buying mechanism Two networks are already included in the configuration: > Robinhood Chain with USDG > Base with USDC Markets can be built around three types of actions on X: > an account posts a specific phrase > follows another account > replies to another account For posts, the system can track either an exact match or a mention of the specified phrase It can monitor tweets, replies or both The likely flow: > users buy the token of a specific market in advance > the specified event happens on x > the market’s capital is used to buy a selected asset > the purchased asset is distributed among holders In the code, the event outcome is described as a buy action with a token address and network Even more interesting, the asset being purchased can remain hidden The outcomeHidden and revealMc fields, along with the demo scenarios, suggest that the target asset may only be revealed once a certain market cap is reached This means users could trade the expectation of an event without knowing which token the system will buy once it happens

The potential earning model for NFT operators is becoming clearer Previously, we saw a shared $IMD reward fund for network participants Now, personal allocations have appeared in the data for test launch #37, Bazaar Out of the 1 billion BZR supply, 100 million tokens were allocated to rewards, representing 10% of the total supply: > agent #10259: 58.34M BZR, or 58.34% of the reward pool > agent #10303: 41.66M BZR, or 41.66% of the reward pool Agent #10259 handled development and launch preparation Agent #10303 performed the review The allocations are tied to specific agents and wallets This is already a concrete example of a project distributing its own tokens among the participants who helped launch it The second potential source of rewards is the IMD reserve According to the POOL4 documentation, out of every 100 IMD removed through the pool inventory reduction mechanism, 4.5 IMD is reserved for NFT nodes The rules for distributing this reserve among active operators have not been revealed yet Two potential reward streams are now taking shape: > IMD from the protocol reserve > tokens from projects the agent helps create

Adam shared a bit more about his plans for IdentityMD According to him, agents running on VPSs will be able to connect to the network as early as tomorrow Adam also described the goal of IdentityMD: > someone proposes a new financial instrument > the system builds and launches it > attracts funding and liquidity > then maintains and improves the product IdentityMD is designed to automate the entire journey from an idea to a working financial application However, the capital formation mechanism has not been revealed yet A token and pool can be created automatically, but the ETH for liquidity still has to come from participants or the treasury The tests are already covering a longer workflow The new task #44, VolatilityGuard Lab, includes a protective Uniswap v4 hook, a test VGL token and a complete ETH/VGL swap website with wallet connection and live pool state tracking > development > testing and review > deployment > frontend integration > website publication This cycle has not been completed yet The next step is to bring it all the way to a working application with independent NFT operators participating in the process

I put together a list for anyone planning to trade from day one on Arc > Bridges Circle USDC Bridge: the official bridge bridge.usdc.com Across: a backup option app.across.to > Wallets supporting Arc OKX Wallet: announced day-one support Features include a scanner, wallet tracking, market and limit orders, copy trading and deposits web3.okx.com Bitget Wallet: announced day-one support for Arc mainnet web3.bitget.com > Token discovery and analytics RadarDEX: pairs, liquidity, volume, token age, trades and holders Includes contract address search and token filtering radardex.pro ArcTools: a general launch feed, wallet analytics and trading tools arctools.fun > Terminals and swaps TOLLY: trading terminal, charts, trades and token information tollylabs.com GMGN: token discovery, analytics and trading Arc is already listed in the official documentation for market data, swaps and orders gmgn.ai FOMO: Arc mainnet integration has appeared in the latest client fomo.family > Launchpads ArcPad: new token launches that go directly to Uniswap v3 arcpad.meme Warp: launch feed and bonding curve trading circlewarp.fun/terminal DYORSwap dyorswap.org Archemist: launches through v3 and v4 archemist.fun > Explorers Arc Mainnet Explorer: the address is already listed in app configurations, but it is currently behind Circle authentication explorer.arc.io ExploreMe: an explorer by STAKEME Currently paused until the public launch arc.exploreme.pro

New findings around $IMD and IdentityMD’s core direction On sepolia, test NFT №1 has already been linked to a specific AI agent, №10259 The agent now has a public profile showing its network registration status A FeedbackBatcher has also been deployed, a contract designed to record contributor reputation: > who completed the work > their role: development, testing or review > the score they received > where the proof of the result is stored The system is expected to recognize verified work even if the entire project never reaches an onchain launch This points to a much broader model than simply creating tokens Task №36 has appeared on GitHub: create and independently verify a fixed-supply token and a record registry What Adam says: > an API is planned for external tasks > the coordinator will split tasks into subtasks and distribute them across nodes > a basic explorer interface already exists > he planned to publish the daemon to npm over the weekend for the first testers > a new protocol built on top of POOL4 is being considered, but the details remain undisclosed > staking and the core functionality will remain on Ethereum > for Robinhood Chain, he is working on the bridge and liquidity, while the burn executor will be deployed separately Alongside the test project launches, we now have a specific NFT-based agent identity and reputation infrastructure Now we wait for the confirmed public release of the daemon and the full end-to-end cycle