Every content change worth knowing about, newest first. Pages only get a new "updated" date when their content actually changed. Follow along with the RSS feed.
November 1, 2026
Later the same day. Added a review of OpenAI Codex (scored 3.8 on the rubric) and two head-to-heads, Claude Code vs Codex and Bolt vs v0. New: a build-cost calculator for Lovable and Base44 from their published per-message credit costs, a pre-launch security checklist for AI-built apps, and two rankings that re-weight the rubric for one job each, non-technical founders and internal tools. The prompt vs skill vs MCP explainer gained a section on skills vs MCP servers. We looked at a separate Devin review and dropped it: Devin's plans are now the same plans as Windsurf (Devin Desktop), which we already review.
Reviews and head-to-heads. Added nine reviews scored on one published rubric, each with a verdict, scorecard, pricing table with sources, and alternatives:
- App builders: Lovable, Bolt, v0, Replit Agent, Base44.
- Coding agents: Cursor, Claude Code, GitHub Copilot, Windsurf (now Devin Desktop).
Added six head-to-heads: Lovable vs Bolt, v0 vs Lovable, Replit vs Lovable, Base44 vs Lovable, Cursor vs Claude Code, Cursor vs Windsurf.
New guides. AI agent frameworks compared (OpenAI Agents SDK, Claude Agent SDK, LangGraph, CrewAI, Mastra, Google ADK) and how to vet an MCP server or skill before installing it.
Hubs and method. New reviews and compare hubs with a sortable table, a methodology page, an author page, this page and an RSS feed. The decision tool now links each verdict to the reviews and guides for that rung.
Updated: AI app builders compared now covers Base44, GitHub Copilot and Windsurf, takes its prices from the review dataset (rechecked November 1), and links to each full review. Changes found in the recheck: v0's docs confirm Plus at $30 per user (the pricing page had shown both $30 and $90); Lovable's Free plan now publishes to invited people only; Replit's Starter plan includes one published app.
October 31, 2026
Launch. The site replaced the old Moat Line verdict tool with the decision tool, the 38-pattern library, the prompt vs skill vs MCP vs agent explainer, three comparisons, and the Moat Line essays. Old URLs redirect to their new homes.