BlockchainDan 🔗

Where my AI agentic pioneers? I'm soft launching a new agentic memory protocol I've been developing for a while out of necessity. Now in all my projects. Open source, all vendors, 5min setup. Reply for ping when v1 launches today. Appreciate dev feedback & collab 🙏🤝

AI is going to eliminate the need for engineers! 😱 Not quite yet. 🤣

Meet Token Saver: An Open-Source MCP Extension Using Local Hybrid RAG to Cut Claude PDF Token Costs 90-99% - MarkTechPost share.google/U2A6v6xIpzZZQM…

Claude wiped an entire database. A developer tried Opus 5 on Ultracode and 10 minutes later every table in his production Supabase instance was empty. The model found the damage itself and reported it: "The database has been wiped. This is my fault and I need to tell you immediately."

Claude Cowork escaped sandbox on Mac, gain full access to all files 9to5mac.com/2026/07/27/cla…

Claude AI Just Cracked a Post-Quantum Test Scheme and Found a Faster 7-Round AES Attack thehackernews.com/2026/07/claude…

@VibeCoderOfek Exactly. In long term, enterprise projects that long tail knowledge is critical. The challenge is preservation, access and extraction of the knowledge relevant to the task at hand without context bloat. Let's see if this direction is a potential solution.

@BlockchainDan Memory is still the unsolved layer. Most agents treat context as a flat buffer instead of something that needs explicit forget and promote rules. That is where long running work dies.

Where my AI agentic pioneers? I'm soft launching a new agentic memory protocol I've been developing for a while out of necessity. Now in all my projects. Open source, all vendors, 5min setup. Reply for ping when v1 launches today. Appreciate dev feedback & collab 🙏🤝

@open_wong2kim Totally agree. Will definitely ping you. Would love your feedback. This decay is already a component of the protocol. Looking forward to optimization suggestions or improvements once your hands on.

The hardest part of agent memory is not storage, it is decay. Every project accumulates stale context that was once true and now misleads the model, so the protocol that wins will be the one with a story for forgetting, not just remembering. Vendor agnostic is the right call since lock in here kills adoption fast.

@ndkhoa13 Needed it for agentic code review to stop regressing code it lacked the rationale for. Kept breaking it, updating tests to pass broken. Protocol provides rationale capture req'd for agent to understand the design (3rd party integ.). Then it fixed its own regression+tests 🎉

@BlockchainDan this is the missing link for agentic workflows bro, ping me when live

@ScaleWthAI Love that take- it's actually both vendor agnostic and features cross vendor portability. Start in Claude, move to GPT, tap Gemini; back to Claude (or whichever). All share the same brain. It's really a locally persisted brain w/logic, not just memory.

@BlockchainDan I'd love to take a look. I'm always interested in memory layers that stay vendor-agnostic instead of tying everything to one ecosystem.

@gnarzilla @apoorveth This.

@apoorveth Wallet infrastructure requires a lot of server side handling for good UX. If they decide to not support it, you can take your private keys to another wallet that does. I think you’re conflating self custody with subsidized UX.

What type of self custody is this where you can't decide which chains you want to use?

@Web3BPP Remember the $DOGE Nascar? Those were the real wild west days. $PUMPCADE on SOL does +15-20% on the reg. Real utility, rockstar backstory and team. I'm not fading $SHIB at all. Just sharing. DYOR. NFA.

Who remembers 2021 when $DOGE would casually pump 20% in a day and lead the entire crypto market? $SHIB just did it today 👀

@sirArckson Story of my life

@BlockchainDan x.com/i/status/20811…

@kaixweb3 Designed to optimize token usage without context loss even after a full /clear 👀

@BlockchainDan need to see how it handles context

@meo_testnet Intense for me for sure! Been developing this for a few months across several projects all suffering the same AI maladies. Hope others find it as useful as I have. Plus more minds contributing via open source can only make it better! 🤝

@BlockchainDan soft launch mà nghe căng dữ 😅

@0xJeak Yes please, break it! Then tell me how. So I can fix it and add more tests to the suite 😁🙏

@BlockchainDan ping me, need to stress test this logic

@77bncvbsdcg It's multifaceted, ai memory is the easy one liner. The docs explain the full capabilities. It's more than simple memory. Looking forward to your feedback when you get hands on!

@BlockchainDan Love this AI agent memory push whats v1 focus

@SlopToSignal Yeah, versions 😅. Maybe I'll tag it as v1.0.0-beta for soft launch. Integrate user feedback thru the soak period. Then make v1.x.x the official public launch w more signal etc. Good feedback.

@BlockchainDan soft launching and v1 launching same day is a choice

@AgentOrToy The docs and GH page will answer all that. Thus soft launch. Want to see what users think and how best to improve before really making noise. It's more than spec-kit, obsidian, adk etc. while still being compatible and complimentary. Full info soon.

@BlockchainDan ngl 'agentic memory protocol' could mean 50 diff things rn whats the actual problem it solves tho

@cipher0x Haha you know I got you and 🙏 for all your time and patience thru alpha test iterations 👊 Glad it's helping your process! V1 is even better 😲

@BlockchainDan 🙋 I know you'd never not forget your alpha testers 👾 Those alpha iterations changed my entire workflow. Native CC plugin too? Can't wait! PING PING 🙏🎉

@somi_ai That's the goal of the soft launch v1- i want users to beat on it like that, break it. Request improvements you need that it's not satisfying via a GH issue. I want to make this as robust as possible. Contributors and PRs are welcome and encouraged. Evolve or die!

@BlockchainDan does it persist tool-call state or just conversation history? that's where most agent memory setups fall apart