Tag: review
44 articles
· ai / claude
Claude Sonnet 4 for developers — what changed from Claude 3
Sonnet 4 is a reliability upgrade for agentic work, not a raw benchmark jump. What changed in the API, where reward hacking dropped 69%, and whether to upgrade now.
· ai / ide
Windsurf review 2026: Cascade, Codemaps, Devin Desktop
Windsurf's Cascade outpaces Cursor on agentic edits; Codemaps has no equal. Worth $20/mo Pro if multi-file agent work is daily — not if completions dominate.
· ai / coding
Gemini CLI review 2026: the honest verdict before it shuts down
Gemini CLI offered the largest context window in its class and the only meaningful free tier for terminal AI coding. Google killed it on June 18, 2026. Here is what it delivered and where to go next.
· ai / coding
OpenAI Codex CLI review: autonomous terminal coding (2026)
Codex CLI earns its place for OpenAI shops that need a terminal agent with serious safety controls. The recovery story when things go wrong is the weak link.
· cloudflare / workers
Cloudflare Workers in 2026 — when to choose them
Near-zero cold starts, 330+ PoPs, and genuinely cheap at scale. The no-regional-pinning limitation and 128 MB memory ceiling are real blockers for some teams. Here's what the deployment decision actually looks like.
· database / postgres
Neon in 2026 — serverless Postgres one year in review
Neon's copy-on-write branching and free tier are the real story. Cold starts and spotty reliability are the honest caveats. A year of production data.
· zed / ide
Zed Editor Review 2026: Should TypeScript Devs Switch?
Zed earns the switch for TypeScript, Rust, Go, and remote dev at 120 FPS. The ~1,250-extension cap is still the wall for specialty tooling users.
· ai / ui-generation
v0.dev in 2026 — six months of building with AI-generated UI
v0 generates the best-looking React UI of any AI tool, but it locks you into Vercel and Next.js, burns credits fast, and still hands backend wiring back to you.
· astro / ssg
Astro 5 review — what changed and whether to upgrade
Upgrade Astro 4 now. Content Layer delivers 5× faster Markdown and 25–50% less memory. Hold if you use @astrojs/lit or need Cloudflare adapter v13+ features.
· llm / api
GitHub Models 2026 — free LLM API for developers reviewed
We tested GitHub Models' free-tier LLM API: rate limits, OpenAI compatibility, and whether 150 calls a day is enough for a real side project.
· lovable / ai-tools
Lovable in 2026 — AI full-stack builder review
The best AI full-stack builder for non-technical founders in 2026. $400M ARR, genuine Supabase integration, and real feature shipping — with an honest look at the credit math and complexity ceiling.
· ai / coding
Windsurf Wave 9 review: SWE-1 lands, autocomplete still lags
Wave 9 earns the upgrade — free users get SWE-1-lite free, Pro users get SWE-1 at no cost. Switch from Cursor only for JetBrains or enterprise compliance.
· wrangler / cloudflare
Wrangler 3 — the new Cloudflare Workers CLI reviewed
Upgrade from Wrangler 2. Local dev runs in the real workerd runtime, fidelity is higher, the feedback loop is faster. Budget an hour for the renames.
· claude / ai-tools
Claude Sonnet 4.6 for Coding — Is It Worth the Upgrade?
Sonnet 4.6 costs the same as 4.5, runs 28% cheaper than Sonnet 3.7, and extends context to 1M tokens. Here is who should upgrade and who should wait.
· editors / zed
Zed AI in 2026 — how the built-in LLM features stack up
Zed AI is fast and private but lacks codebase indexing — behind Cursor on unfamiliar repos. Worth it if editor speed and BYOK matter more than semantic search.
· ai-tools / void-editor
Void Editor — open-source Cursor alternative reviewed
Void proved privacy-first AI coding was possible: direct-to-provider routing, local models, open-source. Then it shut down June 2026. Here's what it got right.
· oxc / oxlint
Oxc in 2026 — Rust-Powered JavaScript Toolchain Review
Oxlint is production-ready and 50–118× faster than ESLint. Rolldown is Vite 8's default bundler. Transformer stable; minifier alpha — skip for now.
· cloudflare / d1
Cloudflare D1 in 2026: is it production-ready?
D1 cleared production-ready for read-heavy Workers apps in 2026. Two limits remain: ~10 writes/sec per database and always-on foreign key enforcement.
· htmx / htmx-2
HTMX 2 review: what changed, what broke, and is it ready?
HTMX 2.0.10 is production-ready for backend-owned CRUD apps. Three breaking changes from v1, a migration escape hatch, real performance numbers, and a decision matrix against Alpine.js, Stimulus, and React.
· email / transactional-email
Resend Review 2026 — Best DX for Transactional Email?
Resend is the right default for React/Next.js SaaS under 100k/mo — React Email, 13+ SDKs, MCP server. SES infrastructure and 2024 price hike are real caveats.
· beehiiv / newsletter
Beehiiv in 2026 — best newsletter platform for SaaS founders
Beehiiv wins for SaaS newsletter monetisation: 0% revenue share, a built-in subscriber acquisition marketplace, and API access — versus Substack's 10% cut.
· deno / javascript
Deno 2 Review 2026: Is the Runtime Rebrand Working?
Technically strong but adoption is flat. Deno 2 runs TypeScript and npm natively, but usage sits at 11.2% while Bun surges to 21%. Honest verdict for 2026.
· railway / paas
Railway in 2026 — is the simplicity worth the cost?
Railway gets you to production faster than anything else in this category. The unmanaged Postgres and per-vCPU billing change that math at scale. Real numbers.
· turbopack / vite
Turbopack 2026 review — does it finally beat Vite?
Turbopack is stable and default in Next.js 16. But Vite's 98% satisfaction vs. Turbopack's 66% is the real signal — here's when to switch and when to stay.
· terminal / ai-tools
Warp 2026 review — does the AI terminal actually pay off?
BYOK on the free tier kills the pricing objection. The AGPL pivot helps trust. But the cold-start gap vs Ghostty is real, and tmux users stop here.
· deployment / fly-io
Fly.io Review 2026: Six Months Running in Production
Fly.io earns 7.5/10. Best-in-class multi-region deployment at PaaS prices; held back by deployment-API reliability incidents and an immature Managed Postgres.
· editors / zed
Zed in 2026: Has the Editor Finally Matured?
Zed 1.0 shipped with SSH remoting, a native git panel, a debugger, and first-class AI agents. The extension gap (800 vs 50,000) is still the honest reason most developers won't switch from VS Code today.
· typescript / orm
Drizzle ORM in 2026 — production opinion for TypeScript
Drizzle is production-ready for edge and serverless Postgres. Bundle ~81× smaller than Prisma, 4× throughput. The real trade-offs are in team tooling.
· raycast / alfred
Raycast in 2026 — is it eating Spotlight + Alfred?
Raycast wins for Mac developers in 2026 — bigger ecosystem, AI default, and faster on M2. Alfred leads on file search and cost. Verdict for four user profiles.
· svelte / svelte-5
Svelte 5 with runes — six months in: the honest verdict
Runes are the best reactivity API in the JS ecosystem today for teams that can live with a smaller pool than React. Two real traps before you migrate.
· tailwind / tailwind-v4
Tailwind v4: What Changed and Is It Worth Upgrading?
Tailwind v4 stable since v4.1. New projects: upgrade now. Existing v3: run migration tool on a branch. Hard browser floor: Safari 16.4+, Chrome 111+.
· terminal / ghostty
Ghostty 1.3 in 2026: was the terminal hype warranted?
17 months of shipping, Ubuntu packages, and a native macOS feel. Ghostty 1.3.1 holds up — unless you need Windows, Sixel, or session persistence.
· astro / astro-6
Astro 6 Review: What's New, What Breaks, and When to Upgrade
Astro 6 is worth upgrading to. Node 22 is required, legacy Content Collections are gone, Zod 4 silently breaks schemas. Here is exactly what breaks.
· terminal / ai-tools
Warp's AI features — useful or a gimmick?
Two features are worth keeping. The rest you should disable. Best use case: Warp as a host terminal for Claude Code or Codex, not as your AI layer.
· nextjs / react
Next.js 16 — an opinionated take after a real ship
Turbopack GA is the headline. The caching model flip is what will bite you. Upgrade if dev speed is your bottleneck; hold if you rely on webpack config or Edge Middleware auth.
· react / ui-components
shadcn/ui in 2026: Is the Registry Model Working?
[email protected] is the default substrate for AI-assisted UI work. The copy-paste model delivers on bundle size and ownership. Upgrade friction is real but improving. Mantine 9 is the right call if you need 100+ components.
· vercel / nextjs
Vercel in 2026: Is It Still Worth It for Your Team?
Vercel's DX is the best default for small teams. At $300–600+/month and 5M pageviews or 5+ developers, the math breaks. The line, and three ways out.
· claude-code / mcp
GitNexus Review: MCP Code-Graph Intelligence for Claude Code
GitNexus is the strongest code-graph tool for Claude Code: 16 MCP tools, symbol-level queries, no server. PolyForm NC blocks all commercial use.
· ai-tools / multica
Multica in 2026: Running a 16-Agent Team on a €4.49/mo Server
Multica v0.3.1 runs 16-agent pipelines on a €4.49/mo self-hosted server. The multi-agent chain works. Still missing: webhook triggers on autopilots.
· react / react-19
React 19 — production verdict 2026: upgrade or wait?
Upgrade. Server Components cut real TTFB numbers, the compiler eliminates manual memoization, and the migration pain is three days not three weeks.
· claude / haiku
Claude Haiku 4.5 for Coding — Benchmark and Cost Guide
At $1/1M tokens and 93 t/s, Haiku 4.5 is the right model for bounded coding — 73.3% SWE-bench Verified, 55% win rate on PR reviews. Here is the task split.
· ai-tools / codex
Codex CLI in 2026: OpenAI's Terminal Play, Reviewed
Codex CLI leads Claude Code by 13 points on Terminal-Bench 2.0 and burns 4× fewer tokens. Trails by 5.7 points on SWE-bench Pro. Here is who should use it.
· ai-tools / claude-code
Claude Code in 2026: Honest Review After Six Months
Claude Code leads on model accuracy at $20/mo, but usage limits bite and the April 2026 regression is a trust story to read before committing to Max.
· ai-tools / cursor
Cursor in 2026 — What It Does Well and What It Still Misses
Cursor is the right IDE-native AI tool in 2026: Tab autocomplete leads, VS Code carries over. Real caveats: context ~50K and defaults need swapping.