Articles tagged "AI Coding"

**Matt Pocock's "grill-me" is an Agent Skill** that fundamentally changes AI coding-agent behavior with just three lines of instruction ([Zenn writeup by ryonakae](https://zenn.dev/ryonakae/articles/8783c6b3ead2cb)). **The core mechanic**: the agent **doesn't generate code immediately** — instead, it asks **sequential, thorough questions until full mutual understanding**, presenting **2–5 options per question**. The article's author reports **18–24 questions per session**, progressively deepening understanding of project scope, dependencies, and design decisions. When the questioning ends, implementation proceeds **frictionlessly** from the established context. **Problem solved**: fundamentally eliminates the **token-and-time waste** of the traditional plan-mode loop ("plan → build → modify plan → rebuild"). **Advantages**: (1) no plan-modify-replan cycles, (2) full human–AI alignment before any code is written, (3) session history serves as natural implementation documentation. **Recommended environment**: Opus 4.6 or Sonnet at medium effort. **Downside**: the extended questioning phase causes real mental fatigue — the author's argument is that it prevents bigger problems downstream. **Positioning**: grill-me is a practical implementation of [Anthropic Claude Skills](https://claude.com/product/claude-code), sitting alongside [Command Code's Design Partnership Mode](../columns/commandcode-taste-1-personalized-coding-agent-2026-07) and [Crit.md's HITL review](../columns/crit-md-local-first-agent-review-2026-07) as another example of the **"HITL assistant, not full autonomy"** family.

On June 2, 2026, OpenAI added a 'Sites' feature to Codex. Users can generate web apps, dashboards, and games inside Codex and instantly share an OpenAI-hosted URL within their workspace. The supported runtime is limited to Cloudflare Worker-compatible ES Modules, and the feature is currently available as a preview for Business and Enterprise plans only. There is no staging environment, and no official statement on public (anonymous) access. While ideal for internal PoC and departmental dashboards, it is too early to adopt as a foundation for customer-facing production services. The unusual structure of Vercel, Replit, and Lovable being both competitors and partners is drawing attention.

Cursor's May 20, 2026 (3.5) update folds the previously-announced Automations feature (March 5, 2026) into the **Agents Window**, adds **multi-repo and no-repo support, five new marketplace templates, and a 7-day 50%-off promo on agent runs for newly created automations**. This column documents the relationship with Background Agents (an orchestration layer above them), all eight trigger types (schedule, GitHub, GitLab, Slack, Linear, Sentry, PagerDuty, webhooks), the billing model under Private / Team Visible / Team Owned scopes, comparisons with Claude Code Routines / GitHub Actions / Devin, and operational caveats for Japanese enterprises.