Articles

Anthropic redesigned Claude Code Projects on September 17 from a folder of chats into one persistent conversation that scopes work, delegates it across parallel cloud threads, and assembles the result — with shared memory across every thread. It’s a real upgrade, but it’s cloud-only for now and Anthropic is upfront that concurrent threads mean concurrent full sessions burning through your usage limits faster.

Elon Musk promised Grok 4.7 in about ten days on September 2. Fifteen days past that, on September 14, he self-graded the still-unshipped model as ‘roughly on par with Opus 5.0, not 5.1’ — a model Anthropic actually shipped July 24. As of September 17, xAI’s own docs still top out at Grok 4.6.

Anthropic collapsed Claude Chat, Cowork, and Design into a single interface on September 16 and added Claude Docs and Claude Slides in beta, calling the whole thing ‘One Claude.’ Claude Code was not touched — a deliberate line between a consumer/knowledge-worker product and a developer-native one.

Spaii Labs launched EasySpecs.ai on Sept 12, a platform that documents a codebase into a knowledge tree, grounds change requests against it, and ships every spec with machine-checkable oracles and human rubrics before an agent writes code. It’s a paid, SaaS bet on the same problem OpenSpec — a free, open-source, 27,000-star YC alum — is already solving with a markdown folder in your repo.

Coder added Claude Code to Agent Relay on Sept 15, letting regulated enterprises run the Claude Code agent loop inside their own cloud, VPC, or on-prem infrastructure while Anthropic still handles billing and inference. Cursor got the same integration first, on Sept 2 — which says more about the market than about either tool.

Z.ai (Zhipu AI) raised roughly $5B on Sept 13 via a ~$2B Hong Kong share placement and a ~$3B zero-coupon convertible bond, its second raise in two months and third since its January 2026 IPO — pushing total public-market capital raised to roughly $12B, funding compute and next-generation models while Anthropic’s own S-1 remains a confidential draft.

Claude Code v2.1.271 (Sept 14) added per-command allowed_domains for Bash, PowerShell, and Monitor in auto mode with sandboxing, plus omitClaudeMd for subagents, fast mode in Remote sessions, and a marked-up pricing multiplier for internal chargeback. None of it got a launch post — it shipped as one changelog among 60-plus line items.

Manifold Security’s Sept 1 GitSpawn disclosure found that Claude Code’s git status subprocess — and separately, its /ultrareview command — execute attacker-controlled git-config settings before any trust prompt appears. Cursor, OpenAI Codex, and Goose all shipped fixes; the /ultrareview path in Claude Code is still unpatched as of v2.1.270 (Sept 12).

Cognition shipped SWE-2 on September 10, Devin’s new engine, post-trained on Moonshot’s 2.8T-parameter Kimi K3 base with three reasoning-effort tiers. Cognition’s own numbers claim near-Fable-5.1 coding quality at a fraction of the cost — but the model underneath is still someone else’s, four months after Cognition’s $48B Series E.

Cursor launched Projects in beta on September 10 — a cloud-resident ‘coordinator agent’ that doesn’t write code itself but directs other agents that do, subscribing to Slack channels, PR activity, and schedules to act without being prompted. Cursor’s own numbers claim 6x the PR throughput for users who adopt it — but the feature is a beta re-implementation of an orchestration model Claude Code has run natively, headless, and in production for over a year.