Independent · Hands-on · Updated weekly

The best AI coding agents,
tested by people who ship.

No sponsored top-10 lists. No vibes. Just honest reviews of the AI tools dev teams actually adopt — Claude Code, Cursor, Codex, and the plugins that make them better.

Featured Review
Field guide · May 24, 2026 · 9 min read

The CLAUDE.md problem in teams of 5+ — what breaks when AI agent rules outgrow one engineer

Solo CLAUDE.md works fine. The wheels come off around five engineers — drift, contradictory rules, silent token spend. A field guide to what fails and the four habits that fix it.

Read the review
From the review
"A single engineer's CLAUDE.md is a personal document. By engineer five it's actively misleading the model. By engineer ten it has become a fossil layer of contradictions nobody dares delete."

More Reviews

Updated weekly
Reference · 10 min

What AI coding agents actually write to your disk in 2026

Eight agents, eight different folders, ~5–20 MB of memory and session telemetry per day. We map exactly what Claude, Codex, Cursor, Gemini, Continue, Copilot, Aider, and OpenClaw leave behind — and what to prune.
Deep dive · 9 min

Aider deep dive — when to pick it over Claude Code in 2026

Aider uses 4.2x fewer tokens than Claude Code on the same tasks, with a 7-point accuracy gap. We break down the architectural trade-off and the three scenarios where Aider is the correct choice.
Analysis · 8 min

Claude Code pricing in 2026 — Pro almost died in April, what that means for your stack

Anthropic briefly removed Claude Code from the $20 Pro plan in April 2026. They reverted in 24 hours, but the pricing pressure is real. Real math on which tier actually pays off.
Guide · 8 min

How to Write a Great CLAUDE.md (with real before/after examples)

A practical guide to writing CLAUDE.md files that actually change Claude Code's behavior — file hierarchy, what belongs where, and concrete before/after rewrites from production codebases.
Update · 7 min

State of AI Coding Agents — May 2026 Update

xAI's Grok Build and Google's Gemini 3.5 Flash both shipped this month. What changed, what didn't, and what it means for your stack.
Roundup · 8 min

Best Claude Code Tools for 2026

Hand-curated 10 tools, MCP servers, and CLI utilities that earn their place in a daily Claude Code workflow.
Comparison · 9 min

Best AI Coding Agents: Claude Code vs Cursor vs Codex

A qualitative comparison with verified 2026 pricing and the actual trade-offs that matter when picking which to run.
Roundup · 6 min

Best MCP Servers for Claude Code

Anthropic maintains 7 official MCP servers and archived 13 others. Here are the 5 we keep installed.
Open source

We build what we review.

AI Memory Reader is our native macOS app for browsing every CLAUDE.md, AGENTS.md, and JSONL session transcript your AI agents leave on disk. Free, open source, audited by us every working day.

macOS · Swift · GPL-3.0

AI Memory Reader

10 AI agents auto-discovered · live file watching · ⌘F find with highlights · ⌘E edit · chunked JSONL viewer · 3 MB universal binary.

View on GitHub
Consulting

Get your CLAUDE.md audited.

A 90-minute review of your Claude Code setup — file hierarchy, agent rules, the three habits eating your tokens. Written diagnosis, one rewritten root file, and five drop-in templates. $299 solo, $799 team.

90-min Zoom · 48h delivery

CLAUDE.md audit

Written diagnosis · rewritten root file · five reusable templates · one follow-up round. First three engagements free as case studies.

See what's included