Skip to content

Recent entries

Leaving the dedicated IDE to live inside your usual editor — Google ships Antigravity to VS Code and JetBrains, and hands enterprises a "spending valve" and audit logs

Google frees its coding agent from a dedicated IDE, shipping it as an extension for VS Code, JetBrains, and Zed. At the same time, it bundles Antigravity into Gemini Enterprise and gives enterprises control through per-project budget caps and central audit logs. Concerns about lock-in and after-the-

Measuring a Counterpart Who Doesn't Understand "Pass/Fail" — Tricentis Ships a Prototype That Scores AI Agents by Probability and Returns a "Ship-or-Not" Verdict

The faster code gets written, the harder it is for testing to keep up. The three prototypes Tricentis unveiled at its conference center on "Aida," a script-free autonomous tester, and "AgentScore," which grades non-deterministic agents by probability and returns a ship-or-not verdict. Who vets the A

Outsourcing an Agent's Ability to "Look Up What It Doesn't Know"──Firecrawl Wires a Code-Specific Search Box to Over 700,000 Primary Sources

The weak spot of coding agents lies in the quality of the "material they have to look things up in." Firecrawl's newly released Developer Index is a search API targeting only primary sources──repositories, issues, PRs, docs, and more──totaling over 700,000. We unpack it, including its self-reported

When the 'Number You See' Didn't Match the Bill──Claude Code v2.1.239 Fixes a Hidden 10% Markup in Cost Display and 'Silent Double-Billing' at Once

Claude Code has released v2.1.239. The estimates in /cost and --max-budget-usd now reflect the 1.1x data-residency markup that had never been shown. It also fixes a bug where responses were silently re-run in non-streaming mode behind a Bedrock proxy──quietly double-billing──bringing your local cost

Making "Conclusion First, Then Silence" a Choice — Claude Code v2.1.237/238 Takes On the AI's Verbosity and Its "Voice That Reverts on Its Own"

Claude Code v2.1.237 ships a built-in "Concise" output style that drops preamble and running commentary. The next day's 238 also fixes a bug where the chosen manner of speaking reverted to the default midway, and shores up the plumbing around unattended operation. It also touches on what conciseness

Ask another session to "tell me when you're free"—Claude Code v2.1.236 tightens the "wait handling" of parallel unattended agents and a secret-file gap at once

Claude Code v2.1.236 adds a "ping me when you're free" notification to another session, automatic goal check-ins, and an environment variable to pin the starting model. It also blocks a .env rename dodge on macOS, tightening the handling of parallel unattended agents and closing a secret-file gap at

Making "Don't Call a Human Every Loop" the Default — Cursor's Cloud Agents Go All-In on "Stay Resident and Run Autonomously" with /goal and Dedicated Machines

Cursor overhauls its cloud agents. It adds always-on residency triggered by events, /goal for long-horizon objectives, subagents that run on dedicated machines, and PR/Slack subscriptions — pushing toward autonomy that "doesn't call a human on every loop." We lay out both the convenience and the con

Making "Pick Up Where You Left Off When the Limit Resets" the Default — Claude Code v2.1.234/235 Tackle Unattended-Run Wait Times and the Gap Between What Approval Screens Show and What They Actually Grant

Claude Code shipped v2.1.234 and 235 on two consecutive days. Together they bring a feature that waits out a usage-limit reset and continues automatically, closures for credential leak paths, and a fix to make approval screens match the permissions actually granted. A look at the groundwork for unat