AI Coding Live News

On November 4 ClaudeDevs, Anthropic's official developer account, said the status line that a background subagent in Claude Code refreshes every 30 seconds used to re-read its whole conversation each time, and is now written by Haiku 5.5 from short notes, so each update costs about 99% less. Anthropic says sessions with background subagents used about 5% less of the limit, giving about 6% more prompts on the same limit. It is live for most Pro, Max and Ultra users and everyone on Team and Enterprise, on v2.1.316 or later.

Claude Code v2.1.302 (Oct 16) makes fork subagents the default in claude -p, the Agent SDK, desktop and VS Code, starts permission-mode-less headless sessions on third-party providers in default mode, and adds CLAUDE_CODE_AUTO_MODE_MODEL. It also flags that standalone Claude Design (claude.ai/design) closes on Dec 14, 2026; design systems must be migrated to Claude Artifacts, while chats and comments are not carried over.

@CompleteSkeptic (Diogo Almeida, TypeSafe; RLHF background) published public design notes on TypeSafe × coding agents (Google Doc), covering routing, context compaction, subagents, skills, and MCP tradeoffs, and invited the community to experiment. The post drew ~930 likes / ~56k views the same day.

Cursor introduced Projects, replacing per-task chats with a coordinator agent in one persistent thread. The agent stays on, proactively manages work with subagents, and can set reminders, run scheduled tasks, follow PRs to fix CI, and watch Slack for bugs. Agents in a project share memory and artifacts such as plans and demos, synced across your device and agent machines. Available in beta and rolling out starting September 10.

Cursor upgraded its Projects multi-agent architecture with native sandbox security probes. The probes continuously analyze code diffs, permissions, and dependencies before and after subagents commit changes, actively thwarting unintended privilege escalation and security regressions.

OpenAI Developers announced clearer Codex usage visibility: break down quota by tasks, subagents, and individual chats so teams can tune workflows. A quoted post points developers to the Analytics tab in the Codex app and recommends updating to the latest version.

Nous Research (@NousResearch) reported that on Sep 2 @Teknium asked Hermes Agent to clean ~1.06M lines of Python; 1,393 parallel subagents finished in 19 hours, shrinking the repo to ~698k lines (−34.4%), collapsing 37 god files to 6, at about $19,302 run cost versus nearly $2M estimated eng hours. Details: nousresearch.com/refactoring-hermes-with-1393-agents.

Cursor launched Projects. Instead of a new chat per task, you work with a coordinator agent in one persistent thread that proactively manages work with subagents and improves over time.

Cursor launched Projects, introducing a persistent orchestrator agent in a unified thread that dispatches specialized subagents, monitors PRs/CI, and retains deep project-level memory over time.

ECC, an Anthropic hackathon-winning Claude Code setup (MIT), is open source: 68 subagents, 286 skills, and 94 commands covering plan → architecture → code → test → security → review → repair → ship. Repo: github.com/affaan-m/ECC.

Marwan Atef (@Marwan_3atef) highlights Speakeasy’s MIT coding-agent runtime Kit: the model sees one compose tool and can shell, edit, retry, and spin ACP subagents (Claude/Codex/Cursor) in a single round trip — about half the tokens of usual harnesses on size-matched tasks. No sandbox; run it inside a trusted boundary.

Latent Space says it spent over 20B tokens putting GPT-6 Astra through real AI-engineering work—model selection/training, labeling and active learning, keeping pipelines saturated, instrumentation/logs, one-shot deploy-and-debug, and commanding/evaling subagents—claiming serious workflows at under about $6 per hour, beyond Blender demos and games.