标签：#t

According to a new report from Semafor, the White House's decision to impose export restrictions on Anthropic's Mythos was driven in part by fears that it had been accessed by a group linked to China. If the Chinese government actually had access to Mythos 5 or Fable 5, it would present a serious national security […]

Originally published at llmkube.com/blog/making-self-hosted-llm-agents-trustworthy . Cross-posted here for the dev.to audience. Running a single local LLM node is a solved problem. You write an InferenceService, the operator schedules it, llama.cpp or MLX serves it, and you get an OpenAI-compatible endpoint. We have been doing that for months. Running a fleet of them is where it stops being easy. My fleet is heterogeneous on purpose: CUDA pods in the cluster, and Apple Silicon Macs sitting off-cluster on the homelab network, each one running two separate agents (one for inference, one for the agentic coding harness). The day I shipped 0.8.4 to that fleet, I learned exactly how it does not scale. I updated each Mac by hand. The control plane had no idea what version any agent was running. And the launchd reload I used to restart an agent was a silent no-op on an already-loaded service, so the old binary kept running while I believed I had updated it. I found that out by hand-inspecting a process tree. Three machines made it annoying. Thirty would make it impossible, and the whole pitch for sovereign, on-prem AI is that you run a lot more than three. So the last stretch of work on LLMKube was not about a faster runtime or a bigger model. It was about making the fleet trustworthy : able to update itself safely, and unable to lie to the control plane about its own state. Here is what that took. Helm and brew for the edge The fix is a new cluster-scoped CRD, AgentRelease , and a self-update path in the agents themselves. You describe the release you want once, the operator rolls it out, and the agents pull and apply it. The design borrows directly from prior art that already solved this for Kubernetes nodes: Rancher's system-upgrade-controller, k0s autopilot's per-platform SHA-256 staging, and Teleport's outbound-only poll model. The properties that make it safe to leave running: Declarative and approved. An AgentRelease names the agent, the version, and the per-platform

Word scrambling is a deceptively simple mechanic. Rearrange the letters of a word, ask someone to restore the original — that's the entire game loop. But underneath that simplicity is a cognitive process that language researchers find genuinely interesting, and that developers building educational tools keep returning to. The Cognitive Mechanics of Unscrambling When a learner attempts to unscramble a word, they're engaging several parallel cognitive processes: pattern recognition (matching letter combinations to phonemes they know), memory retrieval (searching their lexical database), and hypothesis testing (trying a mental arrangement before committing). It's a lightweight version of the same cognitive work that makes retrieval practice so effective in spaced repetition systems. For language learners specifically, this is high-value low-stakes practice. The scrambled form gives enough context to confirm the answer upon success — no ambiguity like a multiple-choice distractor — while requiring genuine active recall. Implementation Considerations for Developers If you're building a word scramble feature into an educational app, a few things matter: Avoiding anagram collisions: "SILENT" → "LISTEN" is a classic example. Your scrambling algorithm needs to detect valid English words in the output and regenerate if it creates a different real word. A dictionary API lookup on the scrambled result handles this. Difficulty calibration: Longer words and words with repeated letters (like "BALLOON") are objectively harder to unscramble. A good difficulty curve starts with 4–5 letter words and increases length progressively. First/last letter anchoring: Keeping the first and last letters in position is a widely used technique to reduce cognitive load. It's psychologically effective — people anchor on word edges more than the interior. Using Existing Tools vs. Building Your Own For most educational content creators and teachers (non-developers), building their own tool isn't feas

In early 2026, Anthropic researchers ran an experiment with 52 junior developers. Half used an AI assistant to learn an unfamiliar Python library. The other half worked without one. Both groups finished the task. But when tested on how well they understood the code they had just written, the AI-assisted group scored 50% on a comprehension quiz - versus 67% for the unassisted group. That 17-percentage-point gap has a name: cognitive debt. It is one of the most important concepts in software engineering right now, and most developers are not paying enough attention to it. What Is Cognitive Debt? Cognitive debt describes the growing gap between the volume of code that exists in a system and the amount that any developer genuinely understands. It is not a new term, but it crystallized across multiple research streams in early 2026. Addy Osmani (Google Chrome) described it as "comprehension debt" - the hidden cost that accumulates when code becomes cheap to generate but understanding still requires deliberate effort. Margaret-Anne Storey (University of Victoria) formalized the concept in a March 2026 arXiv paper, framing it as a team-level problem and extending it into a Triple Debt Model: technical debt in the code, cognitive debt in the people, and intent debt - the missing rationale that both humans and AI agents need to safely work with code. Cognitive Debt vs. Technical Debt These two ideas are easy to conflate, but they are fundamentally different problems. Technical debt lives in the code - it shows up as slow builds, tangled dependencies, and failing tests. Cognitive debt lives in people - it surfaces as an inability to explain, debug, or extend code that the team themselves wrote. The critical difference: technical debt announces itself through friction. Cognitive debt breeds false confidence. Your tests are green, velocity looks fine, and nobody realizes the system is fragile until something breaks in production and the team cannot reason through why. What the

The blast radius of an API key is not "did it leak." It's "if the agent holding it does the wrong thing, how much of your stack goes with it." A secret scanner answers the first question. Nothing in your toolchain answers the second one before an incident. So I wrote 40 lines that do, offline, from the permission metadata you already have. In short: the blast radius of an API key is set by its permissions, not by whether it leaked: scope width × environment isolation × lifetime × revocability. blast_radius.py reads that metadata (never the secret value, never the network) and scores each key 0–100. In my run a broad prod token hit 91/100 CRITICAL; a scoped key, 14/100 CONTAINED. Stdlib, keyless, deterministic. AI disclosure: I wrote blast_radius.py with AI assistance and ran it myself before publishing. Every score in the output blocks below is pasted from a real run on a synthetic fixture I'll show you — no real keys exist in it; every value is a placeholder like sk-FAKE-… . The incidents I cite ($82K, 9 seconds, 2,863 keys, 93%) are other people's , and I link each source next to it. I label which is which. A token that could delete the database, used to manage domains On April 24, 2026, a Cursor agent running Claude Opus 4.6 dropped a production database, and the volume backups with it, in about 9 seconds, on one API call . The team behind PocketOS wrote it up. The agent hit a credential mismatch, went looking, and found a long-lived Railway CLI token sitting in an unrelated file. That token had been minted to manage domains. But it carried blanket authority over the whole Railway account, volumeDelete included. No scope tied to an environment. No read-only mode. Recovery took the weekend ( The New Stack, 27 Apr 2026 ; The Register, 27 Apr 2026 ). Read that again, because the lesson isn't "the agent went rogue." The lesson is the token . A domain-management task does not need volumeDelete . The key was over-scoped before the agent ever touched it. The agent didn'

Hermes Agent already has memory, and that matters. It keeps local context, it improves over time, and it works without forcing you into a cloud service. It also supports several external memory providers. I still built hermes-mempalace , because none of the existing options fit my setup quite right. I wanted something: local-first isolated by Hermes profile verbatim, not just extracted facts easy to inspect on disk simple enough to trust over time That last part is the important one. I did not want a memory layer that turns conversations into an opaque pile of embeddings or summaries you cannot really audit. I wanted actual transcripts, mined into a readable structure, with no hidden server in the middle. Why the existing options were not enough Hermes already gives you a few paths: built-in memory and session context external providers for different use cases enough flexibility to adapt, if you are willing to bend your workflow around them And to be clear, some of those options are good. But ... I run Hermes on a headless machine at home. And I use separate profiles for different contexts. And I do not want conversation content depending on a cloud API or a separate service unless there is a very good reason. So, the best fit had to check a few boxes: [x] no API key [x] no external server [x] no extra runtime I did not already want/install [x] storage isolated by HERMES_HOME [x] memory we can actually read later That let to MemPalace , or https://mempalaceofficial.com/ (hopefully, that's the right one!) What hermes-mempalace does hermes-mempalace wires MemPalace into the Hermes memory provider interface. It follows the same lifecycle as the rest of Hermes memory providers: system_prompt_block() adds a short memory reminder to the prompt. prefetch() can run a MemPalace search before the first model call. sync_turn() buffers completed turns without slowing the chat loop. on_session_end() writes buffered turns to markdown and mines them into the palace. shutdown() flu

Claude Code activity, right where you want it. Discussion | Link

OpenAI launches the Partner Network, investing $150M to help global partners accelerate enterprise AI adoption, deployment, and transformation.

Startups are trying to "ride that SpaceX IPO wave."

I have this vivid memory of walking to pick up my oldest from school in June of 2022. For a variety of reasons, I was in a very bad place mentally. And to make matters worse, it was brutally hot. I was depressed, angry with the world, sunburned, and soaked through with sweat. But as […]

This is a current list of where and when I am scheduled to speak: I’m giving a keynote at Cybernation 2026 in Berlin, Germany, on June 24, 2026. I’m speaking at the Potsdam Conference on National Cybersecurity at the Hasso Plattner Institut in Potsdam, Germany. The event runs June 24–25, 2026, and my talk will be the evening of June 24. I’m participating in a panel discussion at the Austrian Institute for International Affairs in Vienna on Thursday, June 25, 2026. I’m speaking at the Digital Humanism Conference in Vienna, Austria, on Friday, June 26, 2026...

Welcome back to TechCrunch Mobility, your hub for the future of transportation and now, more than ever, how AI is playing a part.

University of Leicester historian thinks Eilmer of Malmesbury saw two different comets: in 1018 and 1066

Hoping to catch some World Cup matches while spending as little money as possible? You have a few options for finding a few days of free streaming, although you may choose to eventually pony up some money. That, or get creative by combining multiple offers to make it through the whole tournament. We found a […]

So you built your stack on a hosted frontier model. Good throughput, clean API, your foreign-national engineers hit the same endpoint as everyone else. Then on June 12 the US government pulled Claude Fable 5 and Mythos 5 offline for the entire planet, three days after launch, and the reason is a compliance gap baked into how these things actually serve traffic. Here's the thing worth understanding as an engineer: the bug was narrow. The takedown wasn't. The gap between those two facts is where every team running a hosted model should be paying attention. What actually triggered it Commerce hit Anthropic with an order barring access to both models by any foreign national, anywhere, inside or outside the US, including Anthropic's own foreign-national staff. The stated trigger was a jailbreak: point the model at a codebase, ask it to find flaws. That's it. Anthropic reviewed the demo and watched it surface a handful of already-known minor vulns, the kind GPT-5.5 and other public models hand you with no bypass at all. So the capability wasn't exotic. It was automated code review on a Tuesday. The reason it went nuclear is the legal layer sitting on top, not the finding itself. The architecture problem: you can't gate on a passport you can't see Walk it through like any other access-control question. The restriction names a class of users: foreign nationals. Every one of them, globally. Now look at what a model API knows about a session at request time. restriction: deny any foreign national, anywhere session metadata: auth token, IP, usage tier NOT in session: verified nationality isolatable set: ∅ only compliant state: serve nobody An API session doesn't carry a verified passport. IP geolocation is trivially defeated by a VPN and tells you location, not citizenship anyway. There's no field in the request that maps to the restricted class. When you can't isolate the users you're forbidden to serve, the only provably-compliant state is serving no one. Off switch. Global.

Six months ago I moved my RAG pipeline from Pinecone to self-hosted Qdrant. My vector search bill went from $210/month to $6.50/month. Same latency. Same recall. Here's exactly how. The Setup My app does document Q&A for legal contracts. The numbers: 5.2 million vectors (1536-dim, OpenAI embeddings) ~800K queries/month P99 latency requirement: < 50ms On Pinecone Serverless, this cost me roughly $210/month — storage plus read units plus write units for daily ingestion of new documents. What I Moved To A single Hetzner CX32 server: 4 vCPU, 8 GB RAM, 80 GB SSD €8.50/month (about $9.20) Qdrant running in Docker Automated daily backups to S3-compatible storage ($0.50/month) Total: ~$10/month. That's a 95% cost reduction. The Migration Was Easier Than Expected bash# Export from Pinecone (I used their scroll API) python export_pinecone.py --index legal-docs --output vectors.jsonl Start Qdrant docker run -d -p 6333:6333 -v ./storage:/qdrant/storage qdrant/qdrant Import python import_qdrant.py --input vectors.jsonl --collection legal-docs The whole migration took an afternoon. The Qdrant Python client is straightforward, and the API is surprisingly similar to Pinecone's. Performance Comparison I ran the same 10,000 test queries against both setups: MetricPinecone ServerlessQdrant Self-HostedP50 latency23ms4msP99 latency89ms12msRecall@100.970.97Monthly cost$210$10 The self-hosted Qdrant is actually faster because the data sits in memory on the same machine. Pinecone Serverless loads data from object storage on demand, which adds cold-start latency. When Self-Hosting Is a Bad Idea I want to be honest about the trade-offs: Don't self-host if: You have zero DevOps experience and no one on the team does You need 99.99% uptime SLA for enterprise customers Your vector count is growing unpredictably (10M one month, 100M the next) You're a team of 1-2 and every hour on infra is an hour not building product Do self-host if: Your scale is predictable (you know roughly how many vectors

Giving an LLM a database connection is one of those ideas that sounds great in a demo and terrifying in production. The agent writes a slightly-wrong query, and now you're explaining to your team why orders is empty. So when I wanted an AI agent (Claude Desktop, in my case) to answer questions about a SQLite database, I didn't want to hand it a read-write connection and hope for the best. I built a small MCP server that gives the agent read-only SQL access — and I made "read-only" mean it, with two independent layers of protection. Here's how it works, and the design decisions that matter. Full source: github.com/skycandykey1/mcp-sqlite-server (MIT). A 30-second primer on MCP The Model Context Protocol (MCP) is an open standard for connecting AI apps to tools and data. An MCP client (Claude Desktop, Claude Code, Cursor, ...) connects to MCP servers that expose three kinds of capability: Tools — functions the model can call ( query , list_tables , ...) Resources — read-only data the model can pull in (a schema, a file) Prompts — reusable prompt templates The Python SDK ships a high-level helper, FastMCP , that turns this into a few decorators. The interesting part isn't the protocol — it's the safety design behind the tools. Design: keep the dangerous part away from the protocol The first decision: the read-only safety logic has zero MCP dependency. It lives in a plain module ( db.py ) that knows nothing about MCP, so I can unit-test it with nothing but the standard library. The server ( server.py ) is a thin wrapper. That separation matters: the part that must never be wrong (write protection) is testable in isolation, without spinning up an MCP client. Two layers of write protection A single guard is a single point of failure. So write protection happens twice, independently. Layer 1 — open the database read-only at the engine level: import sqlite3 def connect ( path : str ) -> sqlite3 . Connection : """ Open a SQLite database in READ-ONLY mode. Any write raises Op

Every backend engineer has seen this happen, you build an application on a relational database like MySQL, handling thousands of concurrent transactions effortlessly. Then, the business asks for a real time analytics dashboard. But when you run an aggregation query over historical data, suddenly the database that effortlessly managed live traffic starts thrashing, evicting your working set, and dragging application performance down. This isn't a tuning problem, a missing index, or a badly written query. It’s a fundamental architectural collision. OLTP (Online Transaction Processing) OLTP encompasses nearly every concurrent digital interaction triggered across a distributed system. A user downloading a PDF, a microservice firing an automatic maintenance log, a comment on a social feed these are all transactions. Data engineers rely on OLTP systems (like MySQL or PostgreSQL) to capture these concurrent streams of interactions for creating , updating and deleting records. The Tree Based In-Place Engine To reliably capture massive volumes of transactions without corrupting data or locking up the application, OLTP systems rely on a highly optimized, row oriented architecture built around the B+ Tree. Because they must provide immediate, atomic updates to existing records, transactional databases manage state through a strict sequence of physical tree traversal and in-memory page mutation: The B+ Tree Indexing: When a transaction reads or updates id: 1, the engine traverses a B+ Tree from the root, through the branch nodes, directly to the specific physical leaf node holding that row. This O(\log n) traversal guarantees a fast, isolated point-lookup. It ensures the application always hits the single version of the row without scanning irrelevant data. The Buffer Pool & In-Place Updates: OLTP systems perform in place updates. The database pulls the exact page containing id: 1 from the physical disk into memory (the Buffer Pool). The specific row is mutated directly in RAM

Two weeks ago Visa invested in Replit. Not for code collaboration. For agentic payments. TechCrunch reported it on May 28: Visa put money into Replit specifically to "power agentic payments for developers." Over 1,000 Visa employees already use Replit for prototyping. Now they're building the pipes for autonomous agents to spend money. Here's why this matters: Visa doesn't make bets on developer tools. They make bets on payment volume. When they invest in agentic payment infrastructure, they're not guessing — they see the transaction data. And the data says autonomous agents are about to move real money. The question for developers: when your agent needs to pay for an API, a cloud instance, or another agent's service, what tooling stack do you actually use? The Stack Nobody Agrees On Right now there's no standard agent payment stack. But a pattern is emerging across the open-source projects shipping on HN: Layer 1: The Authorization Wrapper Before your agent touches money, something needs to say yes or no. Three approaches are competing: Budget Caps — Set a dollar limit per agent, per day, per category. Tools like AgentBudget and RunCycles enforce limits before execution. Simple, but brittle — what happens when your agent hits the cap mid-task? Policy Layers — Define rules: "Agent A can spend up to $50/day on OpenAI, $200/month on AWS, nothing on ad platforms." Tools like Ledge and PaySentry ship policy engines that evaluate every transaction against a rule set. More flexible than caps, but policy management becomes its own problem at scale. Spending Mandates — The agent gets a formal spending authorization with scope, duration, and approver. Nornr takes this approach: before the agent can spend, a human signs off on a mandate document. Most audit-friendly, least autonomous. Layer 2: The Payment Rail Once authorized, the agent needs to actually move money. The options: Rail Best For Limitation Stripe Agent SDK Subscription SaaS, metered APIs Requires merchant accoun

This is a submission for the June Solstice Game Jam . What I Built Solstice Cipher is a small browser puzzle game about the longest day, code-breaking, and the turning point between signal and shadow. The player rotates mirrors to route a solstice beam through every cipher node before landing on the final beacon. Each level is a tiny circuit of light: if the beam misses a cipher gate, the beacon does not unlock. The game is inspired by a few June themes from the challenge prompt: the June solstice and the long arc of daylight light versus darkness turning points Alan Turing, code-breaking, and computational thinking Demo Demo video: watch in browser / direct MP4 Playable game: https://desciple88.github.io/solstice-cipher-devto-game-jam/ Source code: https://github.com/desciple88/solstice-cipher-devto-game-jam How It Works The game is a dependency-free HTML/CSS/JavaScript canvas app. The board is a 6x6 grid. A sunbeam enters from one side of the board, moves in one of four directions, and reflects when it hits a mirror: / turns east to north, south to west, and so on \ turns east to south, north to west, and so on Cipher nodes record whether the beam visited them. A level is solved only when the beam has touched all required cipher nodes and then reaches the beacon. Controls Click or tap a mirror to rotate it. Use Reset or press R to restart the level. Use Next or arrow keys to switch levels. Use Hint or press H if the path gets stuck. Why the Turing Angle I wanted the Alan Turing category to feel like part of the mechanics, not just a label. The player is effectively debugging a simple signal machine: change one reflector, trace the path, see which gates activated, and iterate until the message resolves. It is not an Enigma simulator, but it borrows the feeling of signal routing, symbolic gates, and systematic code-breaking. What I Used HTML CSS JavaScript Canvas 2D ffmpeg for the demo capture AI assistance was used while preparing the implementation and write-up. I