What AI Update Did You Miss Today?

Base44 Enables Multi-File Backend Functions with Shared Code

Base44 backend functions are no longer limited to a single entry.ts file — logic can now span multiple files via relative imports, plus a new base44/shared/ directory for code reused across all functions in a project. The update is aimed at regular app builders, not infrastructure specialists.

Base44

Claude Code Stops Auto-Running /verify and /code-review

Anthropic changed Claude Code's default behavior so that Claude no longer runs the `/verify` and `/code-review` skills on its own initiative; developers must now explicitly invoke `/verify` or `/code-review` when they want them to run. The skills themselves are unchanged — only the automatic triggering was removed. This flips the experience for two different groups of users: those who relied on automatic checks now need to add explicit invocations to their workflow, while those who found the unprompted reviews disruptive get a quieter, on-demand experience.

Claude Code

Base44 Launches Achievements Rewards Program

Base44 launched a new gamified community program where users complete missions to earn points, redeemable for platform credits or merchandise. New missions are promised regularly, and the program is distinct from Base44's existing Partner/Affiliate Program.

Base44

Claude Code Can Now End Abusive or Jailbreak Conversations

Claude Code gained an EndConversation tool, giving Claude the ability to terminate a session outright when facing severe user abuse or persistent jailbreak attempts. The capability mirrors a safety mechanism that has been live on claude.ai since 2025, bringing the CLI in line with Anthropic's consumer product on this front. The same release bundled a large number of security hardenings and bug fixes, including permission-check tightening across Bash, Windows PowerShell, and background sessions.

Claude Code

Cursor in Slack: Plan Previews, Multi-Repo Support, and Cross-Channel Reach

Cursor upgraded its Slack integration so agents share a plan before starting work, added multi-repo environment targeting (with a manual "Switch repository" fallback), and gave the agent the ability to read from and post to other channels/threads across a workspace, alongside a cleaner footer-link UI.

Cursor

Warp: Auto Save for the Built-In Code Editor

Warp introduced an opt-in Auto Save setting for its built-in code editor, mirroring VS Code's "afterDelay" autosave behavior. Changes save roughly 1 second after typing stops, and again on focus loss (honoring the Format on save preference). It suppresses unsaved-changes dialogs and save toasts, and also applies to the code review diff panel.

Warp

Replit: New Tools Pane Consolidates Workspace

Replit introduced a redesigned Tools pane that consolidates Database, Auth, Domains, Secrets, Publishing, Monitoring, Growth, Security Center, and App Storage into a single pinned panel. Previously these tools were scattered across different parts of the workspace, making them harder to discover. The underlying functionality of each tool is unchanged — this update is purely about organization and discoverability, giving users one consistent place to reach every core app-management feature.

Replit

How Anthropic Used Claude Code to Port Bun From Zig to Rust in 11 Days

Anthropic published a detailed account of how Claude Code handled two large-scale code migrations that would previously have taken years. The headline case: Bun creator Jarred Sumner used roughly 50–64 parallel Claude agent workflows to port over a million lines of Bun's codebase from Zig to Rust in about 11 days, spending roughly $165,000 in API costs while passing 100% of Bun's existing test suite. A second case study covered Mike Krieger porting 165,000 lines of Python to TypeScript in a single weekend. Anthropic framed the results around multi-agent orchestration, adversarial review, and automated compile-test-fix loops as a repeatable methodology, not a one-off stunt.

Claude Code

Bolt: Interactive AI Presentations with Bolt Slides

Bolt introduced Bolt Slides, a new way to build presentations where each deck is a real, interactive web app instead of a static file. Users can prompt Bolt to generate decks with live data, 3D visualizations, embedded prototypes, and interactive charts, then share them as a link rather than a PDF. Bolt Slides ships with presenter mode, content-anchored annotations, and deep-linkable slides, and Bolt also open-sourced the underlying codebase so it can be run with other coding agents like Claude Code, Codex, or Cursor.

Bolt

Claude Code Overhauls /fork and /resume for Background Sessions

Anthropic changed how `/fork` works in Claude Code: instead of spawning an in-session subagent, it now copies the current conversation into a brand-new background session with its own row in `claude agents`, letting a side investigation run independently while the original session keeps going. The old in-session forking behavior didn't disappear — it was renamed to `/subtask`. The same release also reworked `/resume` in the agent view so it opens a picker covering every past session, including ones deleted from the list, and reopens the chosen one as a background session.

Claude Code

Figma: Code-Backed Screens Now Bind to Design Variables

Figma updated how code-backed content is imported onto the canvas: colors, typography, and spacing now automatically bind to existing file variables instead of landing as hardcoded values. The change applies across three import paths: copying a Make preview onto the canvas, generating a design from code via the Figma MCP server, and capturing a live webpage with the Chrome extension. Imported frames also arrive with auto layout enabled more consistently, so edits resize and reflow instead of requiring manual cleanup. The result is less rework when bringing coded or legacy interfaces back into a Figma design system.

Figma

Warp: Clickable Hyperlinks in Terminal Output (OSC 8 Support)

Warp added OSC 8 hyperlink escape sequence support, making URLs in terminal output clickable via Cmd+click, closing three long-standing GitHub issues and matching iTerm2, Windows Terminal, WezTerm, and Alacritty.

Warp

Warp: Pinned Tabs Reach Stable Release

Warp has promoted pinned tabs from Preview to Stable, letting users pin individual tabs or entire tab groups so they stay fixed at the front of the tab bar instead of getting buried under new sessions. The capability was one of Warp's longest-standing community requests, tracked in a GitHub issue open since 2022 that collected over 220 reactions. Pinned tabs persist across app restarts and respect tab group boundaries, extending the organizational model Warp introduced with tab groups earlier in 2026.

Warp

Base44 Adds Voice Dictation for Superagent on Mobile

Base44's mobile app now supports speech-to-text input for composing Superagent messages. Base44 confirmed voice recordings aren't stored — only the resulting text transcript is kept. This is distinct from a full voice-conversation mode, which has not shipped yet.

Base44

GitHub Copilot: Security Reviews in the Copilot App

The GitHub Copilot app now supports a /security-review slash command in public preview, letting developers scan their current code changes for high-confidence security vulnerabilities without leaving the app. The scan targets common high-impact issues like injection flaws, cross-site scripting, insecure data handling, path traversal, and weak cryptography, returning a prioritized, actionable list of findings. It's available to Free, Pro, Business, and Enterprise users.

GitHub Copilot