The one thing
01Two harnesses shipped a build that would not work, and fixed it the same night.
Claude Code 2.1.275 went out at 22:33 UTC. Every request failed with 400 … Input tag 'advisor_20260301' when ANTHROPIC_BASE_URL pointed at a proxy or gateway. 2.1.276 landed at 02:12 UTC with that one line as its entire changelog — three hours thirty-nine minutes of a build that could not reach a gateway.12
Cline's desktop app did the same thing four hours apart. 0.0.31, 21:41 UTC, could fail at launch with desktop backend exited before publishing its endpoint: the app and the Cline Hub it starts had been built from different copies of the same code, so the app shut down the Hub it had just launched. 0.0.32, 01:42 UTC, boots both halves from one copy.34 Caught and fixed inside four hours each, and shipped to users first.
Shipped
03`ubuntu-latest` moves to 26.04 between 19 October and 19 November
Ubuntu 26.04 is GA for GitHub Actions on x64 and arm64, and the ubuntu-latest label migrates off 24.04 gradually across that month-long window. GitHub says the image removes tools and changes tool versions, and that the move may break builds depending on either. Two options and a deadline: test against runs-on: ubuntu-26.04 now, or pin ubuntu-24.04 and decide later.5
Cline 4.1.19 stops guessing when to compact
The compaction trigger had compared a character-based estimate of the conversation, around three characters per token, against the model's context limit. Content that tokenizes denser — disassembly, image dumps, minified sources — filled the real window while the estimate stayed under the threshold, compaction never fired, and the turn was squeezed into a handful of output tokens. The trigger now also uses the token count the provider reports.6
Cline desktop 0.0.31 drives a remote machine over SSH
The feature under the broken build: Cline runs agent tools, workspace discovery, Git metadata and session persistence on a remote host while the app stays local, configured under Settings → Remote. Approvals return to the desktop. Host metadata is written at mode 0600 and records an identity-file path, never key contents. Linux x64 and arm64 hosts only, and one day old.4
Promised, not shipped
Codex rust-v0.155.0 ships `/voice` conversations behind `/experimental`, plus Touch ID verification for MCP requests on supported Macs. Available, unevaluated by us.[^7] · openai-python 3.15.0 adds prompt-cache prewarming and managed Responses WebSocket sessions; langchain-typesafe 0.0.1a2 is a twelve-hour-old alpha; Anthropic's Compliance API now returns Claude in Chrome transcripts, Enterprise beta only.[^8][^9][^10]
The conversation
01Open-weight models now run the majority of production tokens.
- Vercel AI Gateway Production Index, September 2026confirmed against the named vendor page, opened and read
Open-weight models ran 56% of gateway tokens in August, up from 7% in December, and took 14% of spend. Price per token fell 23.2% in August, a third consecutive monthly drop; among teams over ten million tokens in both months the median paid 7.6% less. Fable 5 fell from 13.2% of spend to 4.9% while Opus 5, at roughly half the price, rose to 22.5%. Anthropic kept 64 cents of every dollar.11
One gateway's traffic is one gateway's customers, and Vercel sells the routing that makes model-switching cheap, so 56% is a fact about this population rather than about the industry. What it supports is narrower and still useful: teams here stepped down a tier rather than across a lab, while the frontier defended spend by cutting cache reads to $0.25 per MTok.10
Hacker News ran as the community layer: 29 stories cleared 80 points inside the window. The X layers did not run today — the newest file in both digests is dated 2026-09-16 05:30 UTC, roughly 48 hours old, read as background only; the account roster watches 102 accounts. The Reddit digest's newest file is 2026-09-17 05:30 UTC, collected before this window opened, also background only. No practitioner report is printed above.
Skip this
05open-code-review v1.12.5 Ten features, eight of them viewer restyling. On the watchlist, and nothing to decide.[^16]
Vercel's sub-second artifact deployments A speed claim with no measurement of what it replaced.[^17]
"How Cooley is accelerating IPO work with ChatGPT" A vendor-published customer story. No capability, no availability.[^18]
Bonsai 2 27B — "near-lossless compression in a 9x smaller footprint" 351 points on Hacker News; the figures are the vendor's own, on the vendor's own model.[^19]
GitHub's Copilot impact dashboard feature engagement Admin telemetry. Changes a procurement conversation, not an install.[^20]
Everything we saw
6262 candidates scanned · 4 used in this issue — the rest, with the reason each one was left out
| Item | Source | Signal | Call |
|---|---|---|---|
| anthropics/claude-code v2.1.276 | GitHub Releases | single-line changelog, 02:12 UTC | Led the issue |
| anthropics/claude-code v2.1.275 | GitHub Releases | the regression | Printed with 2.1.276 |
| cline/cline desktop-v0.0.32 | GitHub Releases | launch fix, 01:42 UTC | Printed |
| cline/cline desktop-v0.0.31 | GitHub Releases | remote SSH, failed to launch | Printed |
| cline/cline v4.1.19 | GitHub Releases | compaction trigger | Printed |
| Ubuntu 26 generally available and latest migration | GitHub changelog | ubuntu-latest migrates 19 Oct – 19 Nov | Printed |
| AI Gateway Production Index, September 2026 | Vercel blog | 56% of tokens open-weight | Used as the argument; read as one gateway's population |
| openai/codex rust-v0.155.0 | GitHub Releases | /voice, Touch ID for MCP | Promised, not shipped: experimental flag |
| openai/openai-python v3.15.0 | GitHub Releases | prompt-cache prewarming | Promised, not shipped: unmeasured |
| langchain-typesafe 0.0.1a2 | GitHub Releases | initial release | Promised, not shipped: alpha |
| Claude Platform release notes, 18 September | Anthropic docs | Chrome session transcripts in Compliance API | Promised, not shipped: Enterprise beta. Also the source for Fable 5.1 pricing |
| pydantic/pydantic-ai v2.45.0 | GitHub Releases | durable-run fixes | One MCP session per durable run instead of one per unit, and one DynamicToolset resolution per run — real for durable agents, held for space. Yesterday's 2.44.0 advisory call stands. |
| block/goose v1.51.0 | GitHub Releases | eight features | Gateway pairing allowlist and GPT-live support; nothing that forces an upgrade |
| All-Hands-AI/OpenHands v1.20.0 | GitHub Releases | three features | Per-profile secret scoping and saved profiles in automations; no pin change |
| openai/openai-agents-python v0.22.3 | GitHub Releases | nine fixes | Conditional-approval alignment and session concurrency fixes; patch-level |
| vercel/ai 7.0.106, 6.0.286, 5.0.260 | GitHub Releases | thirteen patch fixes | Stream and telemetry fixes across three majors; no API change |
| alibaba/open-code-review v1.12.5 | GitHub Releases | mostly viewer restyling | Named in Skip This |
| Cursor: Cloud Agents and Automations degradation | Cursor status | minor, 4h42m | Longest incident in the window and the one that hits unattended runs; duration computed from the page's own timestamps.[^12] |
| Cursor: Grok Bot degradation | Cursor status | major, 2h06m | Cursor's second incident of the window, again in a third-party model.[^13] |
| OpenAI: elevated errors affecting ChatGPT Work mode | OpenAI status | major, 1h24m | ChatGPT Work mode; no developer decision follows from it.[^14] |
| OpenAI: elevated error rates across API models | OpenAI status | minor, 24m | Elevated errors across API models, short.[^15] |
| GitHub: elevated errors for OpenAI models in Copilot | GitHub status | minor, 48m | Short and minor; the third-party model again |
| Workflow execution protections in GitHub Actions GA | GitHub changelog | GA | Allowlists controlling who can run workflows — real for anyone letting agents open PRs against a shared repo, held for space rather than doubt |
| Agentic CLI customizations in the usage metrics API | GitHub changelog | new API fields | Reporting on skills, custom agents, MCP servers and slash commands; useful to platform teams, changes no install |
| Run Terminal-Bench and other Harbor evals on Vercel Sandbox | Vercel changelog | changelog entry | Worth a look if you already run Harbor; no figures published |
| The skills CLI now supports Notion hosted skills | Vercel changelog | changelog entry | One hosted provider added; narrow |
| Turbo build machines can now be enabled per deployment | Vercel changelog | changelog entry | Build-infrastructure toggle; off beat |
| Sub-second artifact deployments in Vercel CLI | Vercel changelog | changelog entry | Named in Skip This: a speed claim with no measurement |
| Copilot impact dashboard now shows feature engagement | GitHub changelog | GA | Named in Skip This |
| How Cooley is accelerating IPO work with ChatGPT | OpenAI | customer story | Named in Skip This |
| OpenAI models secretly generate instructions to ignore constraints | OpenAI alignment, via Hacker News | 102 points, 27 comments | Sent to Still Developing: dated outside the window and not read in full |
| Bonsai 2 27B: near-lossless compression in a 9x smaller footprint | web, via Hacker News | 351 points, 113 comments | Named in Skip This: vendor benchmark on the vendor's own model |
| How GLM built its own inference infrastructure | web, via Hacker News | 385 points, 265 comments | A good read on serving at scale; nothing a developer installs or pins this week |
| Bend — a language that blocks AI mistakes via proof | web, via Hacker News | 396 points, 194 comments | No version, no release notes, no availability statement we could open |
| Towards Self-Driving Codebases | web, via Hacker News | 109 points, 84 comments | A position piece, not a release |
| Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data | arXiv, via Hacker News | 133 points, 38 comments | The arXiv layer's only candidate above the floor; changes no practice this week |
| Rate limits on GitLab.com are changing | GitLab, via Hacker News | 163 points, 113 comments | Real for CI that hammers GitLab's API, but not an AI-development tool; off beat |
| X account digest, 102 accounts | digests/ai-dev-x-accounts | newest file 2026-09-16 05:30 UTC | Layer did not run today: nothing collected for this window. Read as background only |
| X keyword digest | digests/ai-dev-x | newest file 2026-09-16 05:30 UTC | Layer did not run today |
| Reddit digest | digests/ai-dev-reddit | newest file 2026-09-17 05:30 UTC | Collected before this window opened; background only |
| GitHub trending sweep | github_trending.py | 0 candidates on beat, 0 pending judgement | The rule admitted nothing and retired nothing today; 112 repositories remain under observation |
End of feed. That is everything from the window worth your time.
Next issue tomorrow, 06:00 UTC — and if nothing ships, it will say so in two hundred words.