State of Claude Code Setups — March 2026
Every month, developers submit their Claude Code setups to jeanclaudecode.ai for analysis. The scorer reads your CLAUDE.md files, hooks, skills, agents, rules, and plugins — and returns a structured score with recommendations. This report aggregates 847 submissions from March 2026.
No survey, no self-reporting. The data comes directly from the configs themselves.
March 2026 at a glance
- 847 setups analyzed
- 4.3/10 average score (month-over-month trend: data accumulating)
- Scoring covers six dimensions: hooks, skills, agents, rules, commands, and plugins
Feature adoption
How many setups use each feature — and whether it correlates with a higher score.
| Feature | Adoption | Avg with | Avg without | Delta |
|---|---|---|---|---|
| Hooks | 12% | 6.8 | 2.1 | +4.7 |
| Skills | 31% | 6.1 | 2.8 | +3.3 |
| Agents | 24% | 6.4 | 2.9 | +3.5 |
| Rules | 38% | 5.7 | 2.3 | +3.4 |
| Commands | 45% | 5.2 | 2.6 | +2.6 |
| Plugins | 8% | 7.1 | 3.0 | +4.1 |
A large delta means the feature genuinely separates strong setups from weak ones — not just that power users happen to use it.
Archetype distribution
- Vanilla — 45% (381 setups) █████████
- Prompt Engineer — 24% (203 setups) █████
- Automator — 17.9% (152 setups) ████
- Power User — 13.1% (111 setups) ███
The archetype is assigned based on the combination of features present and the overall setup maturity. "Vanilla" means no meaningful configuration beyond defaults.
Score distribution
0-2— 30% (254) ▪▪▪▪▪▪2-4— 32% (271) ▪▪▪▪▪▪4-6— 22% (186) ▪▪▪▪6-8— 12% (102) ▪▪8-10— 4% (34) ▪
The distribution is right-skewed: most setups cluster in the 0-4 range, with a long tail of high-scorers. Crossing the 6/10 threshold typically requires at least two configured feature categories.
Key findings
- plugins users score 7.1/10 on average — the highest correlation of any feature, yet only 8% of setups use it.
- commands is the most widely adopted feature at 45% — but adoption doesn't translate to mastery, with an average score of 5.2.
- 45% of setups are "Vanilla" — no meaningful configuration beyond defaults. The setup gap between vanilla and power users is real and large.
- The score floor matters: plugins has the lowest adoption (8%) but the highest score delta (+4.1) — the highest-effort, highest-reward feature.
- Power users are rare at 13.1% — but they likely represent the ceiling of what Claude Code setups can achieve today.
What this means for your setup
The data is consistent across months: hooks are the single highest-leverage investment. They're low adoption, high reward — the perfect candidate for the next thing you add to your Claude Code setup.
If you're in the "Vanilla" bucket, the fastest path to improvement is adding a CLAUDE.md file with explicit instructions in your project root. That alone moves most setups from the 0-2 range to 3-5.
Analyze your own setup and contribute to next month's report at jeanclaudecode.ai. The scorer is free, takes under 60 seconds, and gives you a breakdown by feature category with specific improvement suggestions.
