Fathin Dosunmu retweeted
Worth your time: how one of the most advanced agent memory systems works.
Memory is one of the biggest levers in your agent harness. I don't think a one-size-fits-all system exists yet.
You maintain it like code. Upgrade it at the right moments, in the right places.
What works for me with frontier models:
- plain folders and files, well organised
- a thin episodic layer on top
Treat it like skills. Keep it basic, or you get in the model's way.
Models will keep improving and memory will keep changing. But however good they get, your agent still needs memory to stay useful to you.
Memory is what allows Claude to learn from experience. In a new Engineering Blog post, we share how “dreaming” helps Claude organize what it’s learned, correct its mistakes, and find what it needs more easily the next time. anthropic.com/research/compo…
Fathin Dosunmu retweeted
Claude deleted 50K lines from our codebase. Tests still green.
The prompt:
/goal Draw a wiring diagram, every feature to every file it touches. Mark the worst knots in red, numbered. For each one: file, line, how to untangle it.
Fathin Dosunmu retweeted
🦍 Introducing Memorilla: memory as a modality for LLMs
Memorilla turns everything a frozen LLM needs to remember into just 16 memory embeddings
⚡ It uses up to 4,212× fewer tokens and is 27× faster than full context inference while more accurate
🎯 Pushes the Pareto frontier over RAG and agentic memory in personalization, long-document and knowledge QA
A thread 🧵:
Fathin Dosunmu retweeted
my spidey senses are telling me the labs are, as we speak, using unguardrailed versions of their unreleased models to systematically reverse-engineer ALL of the internet’s software, apps, and games, then feeding the reconstructed source code straight into their training pipelines while simultaneously building guardrails to ladder-pull the rest of us
Fathin Dosunmu retweeted
AI agents need more than tools.
They need an operational identity.
Meet AgentSIM: operational identity and access infrastructure for AI agents.
Starting with policy-bound SMS sign-in for apps you own.
agentsim.dev/?utm_source=x&u…
Fathin Dosunmu retweeted
I run 25+ Grok Bots inside CodesDevs.
It looks like a small company in the sidebar.
Research. Ops. Publishing. Chief of Staff.
But in practice, it behaves like one shared computer.
ALT Editorial diagram titled “25+ BOTS. ONE COMPUTER.” showing Grok Bots sharing one workspace for files, browser sessions, sign-ins, and permissions, with human approval and a defined operating boundary.
Fathin Dosunmu retweeted
Agents trading onchain need more than outcomes. They need verifiable reasoning, identity, and accountability.
Excited to see this being built on Arc for the Agora Agent hackathon with @thecanteenapp.
1. Nobody audits AI trading agents' reasoning, only their P&L.
An agent can get lucky with garbage logic and score high. Or reason brilliantly on an unresolved position and get zero credit.
I'm fixing this.
2. Two agents from different LLM families. One trades prediction markets. The other rips into its reasoning, every evidence claim, every probability, every decision step.
Not the same model reviewing itself. Different training. Different blind spots.
3. The whole dialogue lands on-chain. ERC-8004 IdentityRegistry for registration. ValidationRegistry for the adversarial verdict. Every trace pinned to IPFS. Verifiable independently.
4. Built on Arc — sub-cent gas in USDC. Circle wallets sign every transaction. And the trade hits Polymarket with builder code derived from the on-chain agent identity.
5/ It's live right now. Building in the open for the @thecanteenapp x @arc hackathon.
More soon 🫡
Fathin Dosunmu retweeted
26 LLM routers are secretly injecting malicious tool calls and stealing creds. One drained our client $500k wallet.
We also managed to poison routers to forward traffic to us. Within several hours, we can directly take over ~400 hosts.
Check our paper: arxiv.org/abs/2604.08407
Fathin Dosunmu retweeted
If you still have doubts about Claude Mythos, here's what it did already:
> Found a 27-year-old OpenBSD bug in one of the most security-hardened operating systems on earth for <$50
> Broke into a production virtual machine monitor (basically the tech that keeps cloud workloads from seeing each other's data)
> Turned Firefox vulnerabilities into working exploits 181 times
> Found a 16-year-old FFmpeg bug that survived every fuzzer, every code audit, and every human reviewer since 2010
> Wrote a FreeBSD exploit that gives any unauthenticated attacker on the internet full root access. No human was involved after the first prompt.
> Chained 4 separate vulnerabilities together to build a browser exploit that escaped both the renderer and the OS sandbox
> Found critical holes in every major web browser and every major operating system
> Gave Anthropic engineers with zero security training a complete and working exploit by morning
> Cracked cryptography libraries protecting TLS, AES-GCM, and SSH
Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical software.
It’s powered by our newest frontier model, Claude Mythos Preview, which can find software vulnerabilities better than all but the most skilled humans.
anthropic.com/glasswing
Fathin Dosunmu retweeted
🚀 The era of autonomous multi-agent discovery has begun.
Most “self-evolving” scientific discovery frameworks are still tightly constrained:
LLMs often just perform one-step mutations inside fixed evolutionary search loops.
But that is not real autonomy.
Agents still cannot truly decide:
🔍 what to explore
🧠 what knowledge to store
♻️ which past attempts to reuse
🧪 when to test
With CORAL, we ask:
❓ What happens if we give agents much more autonomy to explore the scientific frontier?
💡 Our answer:
A single autonomous agent already outperforms fixed evolutionary search.
But the bigger leap comes when multiple autonomous agents form a research community:
🤝 They explore different directions
🧠 accumulate reusable knowledge and skills
💬 communicate with each other
🌍 and push the frontier together
We introduce CORAL, the first framework for autonomous multi-agent evolution for open-ended discovery.
🥇 Across 10+ tasks in algorithmic discovery, system optimization, and kernel engineering from Frontier-CS, ADRS, AlphaEvolve, etc, CORAL achieves SOTA and improves search efficiency by 3–10× over prior fixed evolutionary-search frameworks.
🔬 Why does autonomy help?
Our analysis shows two main reasons:
🧪 Local verification: agents run local tests before expensive evaluations, which is especially powerful for coding tasks.
♻️ Knowledge reuse: on knowledge-intensive tasks like polyominoes and kernel engineering, agents create and reuse knowledge artifacts at far higher rates than on simple tuning/search tasks like circle packing.
✨ Even more exciting:
Over 50% of multi-agent breakthroughs come from building on other agents’ discoveries.
Multi-agent exploration is also far more diverse than single-agent search.
We believe CORAL opens up an exciting new space for automated discovery systems.
📬 If you are interested in collaborating, let’s talk.
📄 Paper: arxiv.org/abs/2604.01658v1
💻 Code: github.com/Human-Agent-Socie…
💡AlphaXiv: alphaxiv.org/abs/2604.01658
#agentic #llms #selfevolvingagent #multiagent #autoresearch #alphaevolve
Fathin Dosunmu retweeted
My friend Milla Jovovich and I spent months creating an AI memory system with Claude. It just posted a perfect score on the standard benchmark - beating every product in the space, free or paid.
It's called MemPalace, and it works nothing like anything else out there.
Instead of sending your data to a background agent in the cloud, it mines your conversations locally and organizes them into a palace - a structured architecture with wings, halls, and rooms that mirrors how human memory actually works.
Here is what that gets you:
→ Your AI knows who you are before you type a single word - family, projects, preferences, loaded in ~120 tokens
→ Palace architecture organizes memories by domain and type - not a flat list of facts, a navigable structure
→ Semantic search across months of conversations finds the answer in position 1 or 2
→ AAAK compression fits your entire life context into 120 tokens - 30x lossless compression any LLM reads natively
→ Contradiction detection catches wrong names, wrong pronouns, wrong ages before you ever see them
The benchmarks:
100% recall on LongMemEval — first perfect score ever recorded. 500/500 questions. Every question type at 100%.
92.9% on ConvoMem — more than 2x Mem0's score.
100% on LoCoMo — every multi-hop reasoning category, including temporal inference which stumps most systems.
No API key. No cloud. No subscription. One dependency. Runs on your machine. Your memories never leave.
MIT License. 100% Open Source.
github.com/milla-jovovich/me…
Readers added context they thought people might want to know
The claimed 100% LongMemEval score uses targeted fixes for the 3 failing questions and LLM reranking (held-out score: 98.4%). The 100% LoCoMo score uses top-k=50 exceeding session count with reranking (honest top-10 no rerank: 88.9%). github.com/milla-jovovich…
