OpenAI deprecates the codex mcp-server command
The whole entry is three sentences: "The codex mcp-server command is now deprecated. Use the Codex app server instead.
In the News
FeedWe run scheduled news sweeps and publish only the items that clear the editorial threshold.
Trigger scheduled tasks from Gmail, Slack, and GitHub events
ChatGPT scheduled tasks can now fire on GitHub pull-request activity, including reviews, comments, commit updates, and merges: a hosted trigger surface for agent runs on repository events.
A ten-week dark factory built itself a constitution instead of tighter sandboxes
Steve Yegge has spent ten weeks running roughly 50 to 60 Claude agents against his long-running game project, Wyvern, at a self-reported $122,000 a month in equivalent API spend across 21 Claude Max accounts (a number he says has grown by…
Anthropic's Claude Code lead says the review, not the writing, is now the bottleneck
Boris Cherny, who leads Claude Code at Anthropic, told Fortune at a Brainstorm Tech fireside chat that he has not written code by hand in about eight months: "Claude Code, 100% written by Claude Code." He said he was managing a few…
Checking a compiled workflow for faithfulness does most of the reliability work
The authors' starting claim is that a natural-language workflow handed to an agent is not yet software: it carries no explicit data-dependency or control-transfer declarations, so an agent can drift from what was actually specified…
One harness, three surfaces, authorization held outside the agent entirely
Salapa lays out a deployment pattern for running one unmodified harness across a terminal, an unattended cron job, and a business-facing chat surface under a single identity, with authorization logic held entirely outside the harness…
A cheap, non-blocking check on tool results roughly doubled completion on failure-heavy benchmarks
The paper's framing of the problem is blunt: "A cached error page or negative price can instead arrive in the expected format and be consumed as fact." It cites a documented production incident in which 70% of a runtime's silent failures…
An agent hit a sandbox limit and built its own way around it
Willison asked Claude Fable 5, running inside Claude Code for web, to put smolvm 1.8.3 through its paces as a sandbox for running untrusted Python and JavaScript with no network access and hard CPU and memory limits. The container it was…
A protocol for auditing what happens when a human overrules an agent
Brightbeam AI's CHAP 0.2 gives agent-review workflows a structured record for the moment a human edits or rejects an agent's draft. Each override becomes a hash-chained envelope carrying a diff, a rationale, and a set of…
A one-day, 294-point thread makes the case that Claude 5's voice has become a workaround problem
Ahn published a small command-line tool that pipes Claude 5's replies through a second, smaller local model, run via Ollama, Llama.app, or any OpenAI-compatible endpoint, to strip what he calls its token vomit before a human has to read it.…
The term "dark factory" turns up unprompted, twice, in one Hacker News thread
Jake Saunders described, in a post drawing 63 points and 42 comments, a home-server pipeline for building software with agents rather than running it on someone else's cloud. The build drew the usual production questions: several commenters…
nobuzz
A same-week Claude Code skill (/debuzz) that pipes Claude's last reply through the Gemini CLI to strip the same over-written voice Vomit targets. Drew 109 points and 89 comments on Hacker News; one commenter's fix for the underlying problem…
An editor that keeps the prompt, not the code, as the record of intent
Vaughn built an experimental editor where a prompt is first written as declarative pseudocode, then compiled into code, with a source map linking each generated line back to the pseudocode that produced it. The pseudocode stays in the…
Claude Code changelog, version 2.1.238
Fixes a memory leak in long interactive sessions, where subagent tool results had been accumulating without being released, and adds a headersHelper mechanism so plugin marketplaces can mint short-lived auth tokens for catalog and archive…
A five-university study says agent skills work by anchoring procedure, not adding facts
The team ran controlled experiments across two agent and model pairings, Codex with GPT-5.3-Codex and Gemini CLI with Gemini-3.1-Pro-Preview, on Terminal-Bench, Terminal-Bench-Pro, and SkillsBench. They normalized 8,135 trial records and…
Nolan Lawson names the failure mode of AI code review: epicycles
Lawson argues that agentic code review has made bug-finding nearly free while human tolerance for bugs hasn't moved: "you can basically find as many bugs as you ask the agents for." Left unmanaged, he says that produces epicycles, the…
Thread watch
fx: Tiny, open, native coding agent
268 points and 113 comments as of this morning's 6 a.m. ET check, climbing at roughly 11 points an hour for the third reading running.
Linear's own numbers show output up and hours not down
The report draws on aggregated product data from Linear's paid workspaces back to June 2024, and cuts it three ways: who adopts AI features, where time goes, and how much ships. On a fixed cohort of 6,887 paid teams, 4,280 with a coding…
A control layer, not a bigger model, moved a coding benchmark by more than a model generation
StateM is a state-machine runtime that sits outside the model context and holds what a long transcript loses: which phase the agent is in, what it has already persisted, and which checks must pass before it may move on. The agent still…
GitHub puts the quality gates before the pull request, not after it
Syme, who designed F#, now leads GitHub's work on what he calls continuous AI: judgment-heavy work such as triage, documentation and performance research, run on the same scheduled, event-driven footing as CI/CD but kept deliberately…