Cluster · AI Coding
AI coding tools, ranked by what we ship with.
Pair-coders, agentic IDEs, prompt-to-app builders. Every tool tested in a real build before it shows up here. No vendor sheets, no AI-slop benchmarks, no "10 best" filler. The full spread of what we use, what we'd switch to, and what we keep around for one specific job.
For indie hackers, AI-native builders, and solo agencies shipping product, not pitch decks.
Top picks
If you only read three reviews.
The AI pair-coder that ships features, not just snippets. If you live in the terminal.
IDE-native AI editor that wins for visual / UI / multi-model work; Claude Code wins for terminal + MCP + multi-file refactors.
The default writing assistant in 2026. Strong free tier; Pro tier worth $12 for knowledge workers; AI generation features still maturing.
By use-case
Pick by what you're actually trying to do.
You live in tmux and dotfiles. The IDE is a fallback, not the workspace.
You want a screenshot in, a deployed URL out. Code edits later.
Free tier or under $20/mo. Has to ship without a Series A.
Full set
Every ai coding tool we've reviewed.
| Tool | Starting | VTS |
|---|---|---|
| | From $20/mo (Pro) | 92/100 |
| | Free + $20/mo Pro | 88/100 |
| G Grammarly | Free + $12/mo Pro | 86/100 |
| | Free + $20/mo Pro | 82/100 |
| | Free + $30/user Team | 80/100 |
| | Free + $25/mo Pro | 79/100 |
| | Free + $25/mo Pro | 78/100 |
| | Free + $20/mo Core (annual) | 78/100 |
| | Free + $10/mo Pro | 76/100 |
Coming up
What's queued in this cluster.
- →Cursor → Claude Code: when to switch and what breaks
- →The MCP server stack we run inside every Claude Code session
- →Why we stopped using GitHub Copilot for new projects (and still pay for it)
Working drafts. Each one ships with a real test, real numbers, dated update log. No "10 best in 2026" filler.