today's signal · no scroll
the brief
Agents got more real-world agency today. OpenAI’s Codex can now operate your locked Mac from your phone and run long-horizon goals, while plugin sharing makes enterprise rollouts practical. Google demoed Gemma 4 driving an iOS simulator on-device, Anthropic tuned Claude Code’s session model, and Datasette unveiled an extensible agent—rounded out by a parallel-thinking paper and fresh enterprise guardrails from Cloudflare.
If Goal mode isn’t visible, run “codex features enable goals” and restart; a fix is rolling out to make it default.
Heads up: in the Codex app you might have to ask Codex to enable the feature. > Run `codex features enable goals` Then restart the app. We will have a fix rolling out soon to properly enable it by default. x.com/OpenAIDevs/sta…
From your phone, Codex can securely use apps on your Mac even while the screen is off and locked, extending true remote agent execution across devices.
Codex anywhere and everywhere, all the time. Now your Mac doesn’t have to be unlocked for Codex to use your computer. From your phone, Codex can securely use apps on your Mac, even when the screen is off and locked. developers.openai.com/codex/app/comp… pic.x.com/956aAtM3vl
Goal mode is now in the app, IDE, and CLI so you can set milestones and let Codex work autonomously for hours or days across sessions.
3️⃣ Goal mode is now available in the Codex app, IDE extension, and CLI. Goal mode makes Codex more hands-off, letting you set a goal that it can work towards for hours or even days. pic.x.com/OZ18P1YxBf
Business workspaces can distribute and manage internal plugins across teams, enabling reuse of tools, governance, and smoother rollout of custom capabilities.
Share plugins in Codex across teams. Teams can now distribute custom plugins, reuse internal tools, and manage what’s available across their workspace. Available now to Business users. Enterprises can reach out to enable early access. pic.x.com/jqGjq0h4IC
ChatGPT can create, update, and polish editable PowerPoint slides directly inside PowerPoint, bringing AI-assisted authoring into a core office workflow.
Have you ever thought to yourself: I really don't want to make this PowerPoint. Good news: ChatGPT can now create and edit presentations directly in PowerPoint. Build, update, understand, and polish presentations directly in PowerPoint while keeping slides editable. Now in pic.x.com/q3rTN24AHh
Google shows Gemma 4 E4B autonomously navigating an iOS simulator using Argent, underscoring fast-improving on‑device agents for complex software interaction.
We are entering a new era of on-device automation. ✨ Watch Gemma 4 E4B navigate and drive an iOS simulator directly using Argent. Local models can handle complex interactions and software navigation autonomously. pic.x.com/xuXqx3flOD
Release v2.1.147 keeps background agent sessions alive, renames /simplify to /code-review with effort levels, and improves auto-updates—plus a quick 2.1.148 regression fix.
v2.1.147 — What's changed Pinned background sessions (Ctrl+T in claude agents) now stay alive when idle, are restarted in place to apply Claude Code updates, and are shed under memory pressure only after non-pinned sessions Renamed /simplify to /code-review. It now reports correctness bugs at a chosen effort level (e.g., /code-review high); pass --comment to post findings as inline GitHub PR comments. The old cleanup-and-fix behavior has been removed Improved auto-updater: retries transient n...
Cloudflare CASB now integrates Anthropic’s Claude Compliance API so security teams can observe Claude Enterprise usage centrally with policy controls in the Cloudflare dashboard.
Announcing Claude Compliance API support with Cloudflare CASB — Cloudflare now integrates with the Claude Compliance API, so that security teams can monitor Claude Enterprise activity directly in the Cloudflare Dashboard.
Simon Willison ships Datasette Agent, bringing a pluggable conversational assistant to Datasette with tools and plugins for charts and sandboxed commands.
Datasette Agent — <p>We just <a href="https://datasette.io/blog/2026/datasette-agent/">announced the first release of Datasette Agent</a>, a new extensible AI assistant for Datasette. I've been working on my <a href="https://llm.datasette.io/">LLM</a> Python library for just over three years now, and Datasette Agent represents the moment that LLM and <a href="https://datasette.io/">Datasette</a> finally come together. I'm really excited about it!</p> <p>Datasette Agent provides a conversation...
New arXiv paper explores splitting prompts, thinking, and I/O across streams to parallelize LLM workflows, potentially reducing latency for complex multi-step tasks.
Multi-Stream LLMs: new paper on parallelizing/separating prompts, thinking, I/O — Article URL: https://arxiv.org/abs/2605.12460 Comments URL: https://news.ycombinator.com/item?id=48227923 Points: 67 # Comments: 4
swyx shares a hard-won recipe for turning a vibecoded agent MVP into a maintainable, parallelizable repo after 103 commits of automated refactoring.
working on a "take this vibecoded slop app and make it a production-ready, e2e tested, maintainable, parallelizable agent repo" skill. this thing ran for ~16 hours yesterday and made 103 commits all told and i ended up with exactly the same app but instead of fragile mvp it pic.x.com/xyO6qrXxkZ
