AI Coding Live News

Since Oct 21 many developers on X have complained that Claude Code was draining usage limits much faster. On Oct 28 (UTC) Boris Cherny, who leads Claude Code at Anthropic, replied on X that the team reviewed every session ID users shared, found no single root cause and no regression in Claude Code, and is rolling out fixes for a few smaller issues that made some sessions burn tokens faster. Most fast-drain cases traced back to one or two heavy sessions using xhigh effort, /goal, Workflows, Agent Teams or long SDK jobs.

The unofficial-but-tolerated Claude Code Changelog bot (@ClaudeCodeLog) posted 2.1.274 with 108 CLI changes. Highlights: a visible critical-memory warning with safe free/restart steps; sessions no longer loop forever on unexpected tool_use_id 400s (transcripts self-heal when possible, else a clear /rewind error ends the loop); MCP 403 insufficient_scope now names missing permissions and points to /mcp re-auth; CLAUDE_CODE_MCP_STARTUP_WAIT_MS bounds how long the first non-interactive turn waits for MCP servers. Many more MCP/stream, /goal resume, self-hosted runner 401, VS Code, and cloud session/routine fixes. Source: anthropics/claude-code CHANGELOG #21274.

blankspeaker (@blankspeaker) shipped Grok Build v1.0.22 (2026-09-07): desktop-app tools via a dedicated first-party MCP; built-in agent tools take precedence over user MCP servers on name collisions; remote-control pickers can group servers by device, plus subagent/skills/compact/goal fixes.

GitHub Copilot App weekly recap (Aug 29–Sep 5): 1 release and 237 improvements. Highlights include /goal for persistent autopilot objectives in local sessions, MCP install via deep link with a prefilled review form, Power BI MCP in the catalog (hosted remote and local), and configurable reasoning effort for code-review agents. Notes: github.com/github/app/releases.

Cursor shipped a cloud-agent update focused on long-running, event-driven work. Cloud agents can pick up tasks from events, hold a goal until it is met, and stay on course through long sessions. They can monitor pull requests, watch a Slack thread, or run scheduled tasks, and they automatically subscribe to PRs they create and drive them to completion. Any skill can be pinned as a Custom Mode from the slash menu with Option-Enter or Alt-Enter, acting as an always-on skill. Subagents can now run on their own virtual machines, each with an isolated copy of the project, so they can test the parent agent's changes in a fresh environment or swarm independent fixes. Steering waits for the next tool call instead of cutting the agent off mid-action. A new /goal command assigns a long-lived objective. The full changelog is dated August 19, 2026. Together these changes move Cursor from in-editor chat toward an orchestration layer for isolated, interruptible coding agents.

Cursor updated its cloud agent infrastructure with persistent /goal execution, automated PR event subscriptions, isolated subagent VMs, and non-blocking human-in-the-loop steering.

Cursor updated its cloud agents with event-driven execution, persistent /goal tracking across PRs and Slack, and isolated VM subagents for clean parallel debugging without context drift.

Cursor shipped major cloud-agent enhancements including persistent /goal objectives, isolated VM sandboxes for parallel subagent execution, and automated event-driven PR resolution.

Cursor upgraded its cloud agents to persist long-lived objectives via /goal commands, auto-subscribe to PR discussions and CI events, and spawn subagents inside isolated VM sandboxes. Each subagent runs on a fresh repository clone for parallel testing without dirtying the working branch.

Cursor improved cloud agents so they pick up work from events and hold a goal until it is met. /goal sets a long-lived objective, subagents can run on isolated VMs with their own project copies, and cloud agents auto-subscribe to PRs they open and drive them to completion.

Cursor upgraded cloud agents: /goal holds a long-lived objective until it is done; subagents can run on their own VMs with isolated project copies to test parent changes or swarm independent fixes; cloud agents subscribe to PRs they open and drive them to completion, and can watch Slack threads or run scheduled jobs. Steering now waits for the next tool call instead of cutting the agent mid-action. Any skill can be pinned as a Custom Mode (always-on).