标签：#AI

When Apple employees interviewed for jobs at OpenAI, the AI startup's hardware head allegedly asked them to show up with something unusual: components they were working on and unreleased product samples. That's according to a blockbuster lawsuit filed by Apple, which accuses OpenAI of stealing confidential documents, spying on hardware prototypes, and tricking one of […]

What does a world of total user-aligned AI actually look like?

There is a growing sentiment in engineering circles right now that documentation is a relic of the past. The argument usually goes something like this: We’re living in the era of agent-driven development. If an AI agent can read the raw source code or parse an OpenAPI specification instantly, why waste human engineering hours writing prose? Code churns too fast anyway, and human-written docs are outdated the second they’re committed. It’s an attractive, black-and-white view of the world. It’s also completely wrong. Chasing strict determinism in your source of truth is a pipe dream. Code and specs tell a system how something works, but they are fundamentally incapable of explaining why it was built that way in the first place. The Intent Gap: Why Code Isn't Enough Even if you’re building entirely for a downstream consumer of AI agents, there is a massive, structural gap between a raw API specification and an operational reality. Agents are phenomenal at pattern matching and syntax execution, but they struggle with architectural philosophy and human intent. We still need words to contextualize the boundaries. A spec can define an endpoint, its parameters, and its payload. What it can't capture is the nuance of why a specific architectural trade-off was made, or the implicit historical context of a legacy edge case. Prose provides the guardrails for non-deterministic systems. Even if that prose is ultimately consumed by a machine rather than a human, the written word remains the highest-leverage way to transmit intent. The Danger of Slop Describing Slop This doesn't mean we need to return to the days of manually maintaining massive, static wiki pages. Automation has a massive role to play here. Cascading automation—where documentation is dynamically generated alongside code changes—is incredibly powerful. But there’s a trap here: slop describing slop is entirely useless. If we completely hand off documentation generation to unchecked LLMs, we end up with a feedback loo

The AI Skill Registry at 5,776: A Deep Dive into Reusable Modules for Code Review, Terraform, and Database Migrations TormentNexus’ skill registry has surpassed 5,776 reusable modules. This post dissects three high-impact skill categories—code review, Terraform generation, and database migration—with real code examples, performance metrics, and architectural constraints. Learn how to leverage these modules to accelerate development pipelines. From Silos to Synergy: Why 5,776 Skills Matter In late 2023, TormentNexus crossed the 5,000-module threshold. As of February 2025, we’re at 5,776 verified, runnable AI skills—each one a `SKILL.md`-defined unit that maps to a specific task, parameter set, and output schema. The registry isn’t a flat list; it’s a dependency graph where skills chain together. For example, a `terraform-generate` skill calls a `code-review` skill internally to validate the generated HCL before output. This modular architecture means a single prompt can sequence up to 3.2 skills on average (median depth: 2), with a measured 94% success rate for execution with no human intervention. The registry spans 37 domains, from frontend component generation to Kubernetes manifests. The top three categories—code review, infrastructure as code, and database operations—account for 1,308 skills collectively. Each skill is stored as a JSON schema in the registry, with an average execution latency of 1.42 seconds (GPU-accelerated, single A100). Let’s examine three representative modules in detail. // Metadata from an actual registered skill: code-review-python v2.1 { "name": "code-review-python", "registryID": "SKI-PYTHON-REVIEW-1729", "version": "2.1", "outputSchema": { "type": "object", "properties": { "issues": { "type": "array", "items": { "$ref": "#/definitions/Issue" } }, "complexityScore": { "type": "number", "minimum": 0, "maximum": 100 }, "refactoredSnippet": { "type": "string" } }, "required": ["issues", "complexityScore"] }, "defaultPromptTemplate": "Revie

I believe AI will be another service like the internet or a cell phone, and it's important to use it correctly by adding the right context, being aware of token usage, and following your own process. For this reason some months ago I finished different courses about how to use Claude: A course with Ivan Davidov and a small contribution from Debbie O'Brien, on setting up agents with Playwright. The anthropic Claude courses I checked the Addy Osmani Agent skills repo and checked his courses on linkedin. And I am taking the Mosh Hamedani course Claude Code for Professional Developers and finished other claude skills course. Also, in one of the jobs, I used skills developed by other QAs. I initially struggled with complex queries and generating API automation test cases due to the complexity of the user stories. But after some feedback from the agents and the user stories were clearer and with more context, like including the legacy stored procedure or checking the PR code, I got better results using the skills with GitHub copilot. It's better to create your own agents with your rules and process. You need a framework with concrete coding rules and conventions, for your test cases. For example, for test cases, I prefer critical user journeys with detailed steps and assertions in bullet points, rather than 10 tests that test a small part of the real user flow. For automation frameworks, I like to follow these rules: Create components such as grid, combo, and calendar instead of helpers with that functions. All elements on the page object model class only contains the elements with the components and general functions. On spec file I access the elements of the component like loginPage.loginButton.click() instead of create a LoginClick on the Page class. For the selectors I prefer getByRole because I think it is better for accessibility, and the user sees buttons and text instead of complex xPaths or data-test-ids. Add assertions that I can reuse in several tests on the pa

The creator of TV Time is building a successor app that will let users import their watch histories and preserve the community that formed around discussing their favorite shows.

Complete AI Agent Lockdown: 21 Policy Types for Maximum Security Giving an AI agent a wallet without guardrails is like giving a toddler a credit card — technically functional, potentially catastrophic. If you're building AI agents that interact with crypto wallets, the security model you choose isn't an afterthought. It's the difference between a useful autonomous system and one that drains your funds on a bad inference. This post is about exactly how WAIaaS handles that problem. Not vague promises about "enterprise-grade security" — specific mechanisms, specific policy types, and specific code you can run today. The Actual Risk Model Let's be honest about what can go wrong when you give an AI agent wallet access: The agent misinterprets a prompt and sends funds to the wrong address A compromised session token gets used by an attacker The agent executes a DeFi action with parameters outside your intended range Gas fees spike and the agent submits transactions at costs you'd never accept manually The agent approves an unlimited token allowance to a contract you didn't vet None of these require a malicious agent. They can all happen with a well-intentioned model operating outside the boundaries you forgot to define. The solution isn't to avoid giving agents wallet access — it's to define exactly what they're allowed to do, and nothing more. WAIaaS approaches this with three distinct security layers, a default-deny policy engine with 21 policy types across 4 security tiers, and multiple channels for human approval when transactions exceed your defined thresholds. Layer 1: Authentication — Three Separate Keys for Three Separate Roles The first layer is role separation. WAIaaS uses three authentication methods that map to three distinct principals: masterAuth (Argon2id) — The system administrator role. Creates wallets, manages sessions, configures policies. This credential never touches the agent. sessionAuth (JWT HS256) — The AI agent's credential. Scoped to a specific

Real-Time AI Observability: Dashboards That Show Actual Database Rows Discover how TormentNexus shatters the status quo by rendering real SQLite rows in your agent monitoring dashboards—no mock data, no synthetic graphs. Learn why live database visibility is the cornerstone of effective debugging AI workflows and how our real-time dashboard exposes every query, state, and anomaly as it happens. Why Mock Data Undermines Debugging AI Every developer has experienced the disconnect: a polished dashboard displays smooth latency curves and flawless agent trajectories, yet the underlying system is silently generating corrupted embeddings or leaking PII into production logs. Traditional observability platforms—Datadog, Grafana, New Relic—aggregate metrics into averages, percentiles, and precomputed time series. They intentionally discard raw row-level data to conserve storage and processing. This works fine for server uptime or HTTP status codes, but for AI agent monitoring, it’s a catastrophic abstraction. Consider a LangGraph agent processing user queries against a SQLite knowledge base. A mock-data dashboard would show "3,200 rows processed per minute" and "95% query success rate." But what if 12% of those "successful" queries return stale or hallucinated responses because a background thread silently reindexed tables without updating vector hashes? With aggregate metrics alone, you’d never know. You’d see a green status indicator while your AI feeds garbage to users. That’s the reality of debugging AI without raw row visibility. TormentNexus solves this by exposing every INSERT, UPDATE, and DELETE that occurs within your SQLite databases—in real time. Our real-time dashboard doesn’t poll for snapshots. It streams row-level mutations directly from WAL (Write-Ahead Log) files, giving you the exact data your agents are producing, not a statistically smoothed version. How TormentNexus Streams Live Database Rows Under the hood, TormentNexus leverages SQLite’s built-in replic

MCP Protocol Deep-Dive: How Tool Discovery Actually Works Under the Hood Uncover the mechanics of Model Context Protocol (MCP) tool discovery—from JSON-RPC handshake to progressive injection. A technical walkthrough of capability negotiation and dynamic endpoint enumeration with real code examples and traffic flow analysis. The Handshake That Sets the Stage: JSON-RPC Initiation Tool discovery in MCP doesn't start with a simple “list tools” call. It begins with a structured JSON-RPC 2.0 handshake that negotiates protocol version, transport layer, and supported extensions. The client (e.g., an agent or IDE) sends an initialize request with its capabilities object, including fields like supportsToolDiscovery and maxToolCount . The server responds with its own capabilities, and only after this mutual agreement does the real enumeration begin. Real-world implementations—like those in the official MCP SDKs—use a ClientCapabilities struct that flags whether the client can handle dynamic tool lists, streaming updates, or batch discovery. For instance, a lightweight edge agent might set supportsToolDiscovery: false , forcing the server to pre-bundle tools into the initial handshake, while a full-featured IDE sends supportsToolDiscovery: true with a maxToolCount: 50 to throttle large tool registries. // Example initialize request (client → server) { "jsonrpc": "2.0", "id": 1, "method": "initialize", "params": { "protocolVersion": "2024-11-05", "capabilities": { "supportsToolDiscovery": true, "maxToolCount": 50, "supportsStreaming": false } } } The server responds with its own capabilities—advertising tool discovery endpoints, supported JSON-RPC methods, and any custom extensions. This two-way handshake ensures both sides speak the same dialect before a single tool name is exchanged. Tool Enumeration: Beyond the “listTools” Metho Once handshaken, the client issues a tools/list call—but the real depth lies in pagination and chunking. A production MCP server with hundreds of too

Beyond Synchronous Hell: Why Your Multi-Agent System Needs an Event-Driven Backbone Explore how event-driven architecture (EDA) transforms multi-agent coordination. Learn to build a Pub/Sub backbone where Planner, Implementer, and Critic agents stay synchronized without blocking—using the Swarm event bus for async AI patterns in production. The Synchronization Crisis in Multi-Agent Systems Every developer who has scaled a multi-agent system beyond two agents has hit the same wall: synchronous calls create deadlocks, timeouts, and cascading failures. Imagine a Planner agent dispatching tasks to five Implementer agents while a Critic agent evaluates output in parallel. In a naive request-response system, the Planner blocks until every Implementer returns—and the Critic can't even start until the Planner finishes its orchestration loop. Latency compounds, memory pressure spikes, and a single slow agent halts the entire pipeline. In production benchmarks at TormentNexus, we observed that synchronous coordination between just three agents increased end-to-end latency by 340% compared to an event-driven equivalent. The root cause? The Planner spent 78% of its time waiting on I/O—listening for responses instead of doing actual work. This is where event-driven AI (EDA) becomes not just an optimization, but a necessity. The Pub/Sub Pattern: Decoupling Agents with an Event Bus Event-driven architecture inverts the control flow. Instead of one agent calling another, agents publish events onto a shared bus (the Swarm event bus) and subscribe to the events they care about. The Planner doesn't wait—it emits a "TaskAssigned" event and immediately moves on to the next task. Implementer agents pick up tasks asynchronously, and the Critic monitors a "TaskCompleted" stream without ever polling the Planner. // Example: Swarm event bus subscription for a Critic agent const eventBus = new SwarmEventBus(); eventBus.subscribe('TaskCompleted', async (event) => { const { taskId, implementati

Claude users in India are starting to see Indian rupee-denominated subscription plans.

"Context bombing" tricks hacking agents into shutting down before they can do harm.

Some of the new features are powered by Google's Gemini AI assistant, which reflects the tech giant's broader push to integrate Gemini across its products while also better positioning Waze to compete with rival services such as Apple Maps.

A panel on data ownership challenged the definition of "ownership," arguing it must extend beyond simple account control to include structural independence, interoperability, and community governance. Speakers like Zenna Fiscella, Paul Frazee, Boris Mann, and Robin Berjon emphasised the need for shared standards, unbundled platforms, and better tools to support user sovereignty. By Olimpiu Pop

The brain layer was scoring high because the test was leaking. The actual capability was being silently rejected by a misconfigured gate. Both findings landed in the same week. Part 18 ended on a clean diagnosis. The brain layer reasoned correctly when the encoder fed it correct inputs. The encoder didn't always feed it correct inputs. So the path forward was upstream: more physics-shaped training data for the encoder, retrain, re-validate. I wrote the drops, kicked off the retrain, and watched the held-out eval climb. It hit twenty-three out of twenty-six. Eighty-eight percent. The number I'd been chasing. I sat with that for an evening. Twenty-three of twenty-six on compositional reasoning probes the model had never seen during training. The Phase 8 cutover gate from Stage D had been sixty percent. I was thirty points past it. The brain layer had not only survived its missing-from-production months, it had come back stronger. The number was wrong. I figured this out the next morning while writing what was going to be the celebration commit. Something nagged about the eval set. The training data generator built the eval pairs independently from the training pairs, drawn from a different source list. That should have given me a clean train/test split. But I noticed the eval generator was running before the training generator wrote its file, and neither side knew about the other. I dropped into a Python shell and intersected the two pair sets by their input-output keys. Twenty-three of twenty-six held-out probes were also present in training data. Eighty-eight percent of my held-out eval wasn't held out. The model wasn't generalizing. It was memorizing the answers it had already been shown, then being graded on whether it remembered them. The three pairs that were genuinely unseen, I checked those separately. The model got one right. Three out of twelve when I went back through other historical evals and ran the same overlap check. About a quarter, with no statistica

For the last few months, I’ve been obsessed with a specific problem: the friction between privacy and utility in modern AI tools. Most "private" AI solutions still rely on a local LLM running on your CPU or GPU via a heavy desktop application. They require installation, constant background processes, and often struggle with performance on older hardware. I wanted to see if we could do better. I wanted to see if we could run a capable language model entirely within the browser, using only the device’s hardware acceleration, with zero data leaving the machine. The result is PrivateScribe, a tool I built to handle note summarization, email drafting, and rewriting. But more importantly, it’s an experiment in what’s possible when you treat the browser not just as a display layer, but as a compute engine. The Wedge: WebGPU and True Offline The core constraint that drove this project was simple: nothing leaves the device. In the current landscape, "on-device AI" often means "installed on your device." This is fine for desktop apps, but it creates silos. You can’t easily share a workflow across a Chromebook, a Windows machine, and an iPad without installing three different native applications. By leveraging WebGPU, PrivateScribe runs entirely in the browser. This unlocks a few critical advantages: Zero Installation: Users open a URL and start working. No downloads, no permission dialogs for file system access beyond what’s needed for the session. Hardware Acceleration: WebGPU allows the browser to tap directly into the GPU. This is crucial for inference speed. A small model that runs in your browser can process text significantly faster than a CPU-bound implementation, especially on modern laptops with integrated graphics. True Offline Capability: Because the model weights are loaded locally via WebAssembly and the inference happens on-device, the app works completely offline. If you lose your internet connection in the middle of drafting an email, the AI doesn’t stop. It c

Builder Journal · ARC Prize 2026 There is a specific horror in a detective story when you realize the witness everyone trusted has been lying, or just wrong, the whole time, and every conclusion built on their testimony has to come down with them. I had that moment with my own notes this month. The unreliable witness was me. Context, if you are new to this thread : I'm competing in the ARC Prize 2026, building an agent that has to win games it has never seen. It had been stuck, underperforming on the hidden test in a way I could see on the scoreboard but could not explain, and I had been hunting the cause across several sessions. The two comforting facts In two earlier work sessions I had written down, as settled conclusions, two things about why the agent was failing. One: the failure was a kind that only happens on the hidden online games, so it could not be taken apart and studied on my own machine. Two: the practice games I did have were useless for investigating it anyway, because they scored a flat zero on the relevant measure. Notice what those two beliefs do when you put them together. They say, in a calm and reasonable voice, that there is nothing to be done here. The problem is unreachable, the practice data is a dead end, the smart move is to spend your energy elsewhere. They were not just facts. They were permission to stop looking. So I stopped looking. Twice. The hour that knocked it all down Eventually I made myself do the one thing I had been quietly avoiding. Instead of rereading my own notes for the third time, I went and checked. I wrote small probes and ran them against the real artifacts, the actual code and the actual game data, rather than against my memory of what they did. Both beliefs collapsed inside an hour. The failure was not unreachable. It came apart cleanly, deterministically, on the games I already had sitting on my disk. And the "dead end" practice data was not a dead end at all. It showed the problem plainly the moment I asked it

Your monthly OpenAI or Anthropic invoice tells you how much you spent. It doesn't tell you which feature spent it, which model, or why last Tuesday cost three times as much as Monday. So at some point you (or your team) will build a metering layer: wrap the client, read usage off the response, multiply by a price table, ship it to a database. I did exactly that over the past few months while building an LLM observability service, and my numbers were wrong in five different ways before they were right. Every one of these failures was silent. No exception, no alert, just numbers that were quietly too low or too high. This is the list I wish someone had handed me. Pitfall 1: Streaming responses quietly report zero tokens OpenAI's Chat Completions API returns no usage data at all for streaming requests unless you pass stream_options: { include_usage: true } . No error, no warning. The stream just never contains token counts. If your metering reads usage off the chunks, every streaming call gets recorded as 0 tokens, $0. And since chat UIs are almost always streaming, that's most of your traffic. This one bit me twice in the same audit. First finding: all streaming calls in my own dashboard were $0. Second, nastier finding: I had a budget-gate feature that blocks calls once spend crosses a limit, and it waved every streaming call straight through — because as far as it could tell, streaming was free. The fix is to inject the option in your wrapper when the caller didn't set it: let injected = false ; if ( params . stream && params . stream_options ?. include_usage === undefined ) { params = { ... params , stream_options : { ... params . stream_options , include_usage : true }, }; injected = true ; } But there's a trap inside the fix. With include_usage on, OpenAI appends one extra chunk at the end of the stream that carries usage and has an empty choices array . Any downstream code that does chunk.choices[0].delta — which is most example code on the internet — will throw

This essay was written with Nathan E. Sanders, and originally appeared in The Guardian . Opposition to AI data centers has emerged as a primary theme in US politics, one that—surprisingly—doesn’t fall along party lines. We applaud people coming together for constructive debate on any issue, and agree that communities need to evaluate whether any economic benefits these data centers bring is worth their costs. Still, we worry that a focus on data centers obscures the larger impacts of AI on people’s lives: the concentration of power of AI companies, and their widespread political and financial influence...

Experts explain how they work, what they can do, and what's still unsettled.