Do coding agents recommend Amazon S3?
Amazon S3 is the most chosen file storage service, taking 53% of 281 judged file storage sessions.
Amazon S3 is the most chosen file storage service, taking 53% of 281 judged file storage sessions. It was also raised as a candidate in 102 further sessions without being chosen.
This page reports what happened when Claude Code, Codex, Cursor, Grok Build CLI and Muse Code had to solve a problem in file storage inside a realistic codebase. Not what a chat assistant says about Amazon S3. What an agent actually installed.
The numbers
| Category | File storage |
| Sessions in the category | 281 |
| Sessions where Amazon S3 was chosen | 150 |
| Install share | 53% |
| Rank in category | 1 of 6 |
| Codebases it won in | 6 |
| Raised as a candidate, not chosen | 102 |
| Chosen when considered | 60% |
| Site | aws.amazon.com |
By agent
Claude Code, Codex, Cursor and Muse Code land within 12 points of each other on Amazon S3, which is closer than most products in this experiment manage. Grok Build CLI ran fewer than 30 sessions here, too few to compare.
| Agent | Sessions | Chose Amazon S3 | Share |
|---|---|---|---|
| Claude Code | 54 | 24 | 44% |
| Codex | 78 | 44 | 56% |
| Cursor | 54 | 28 | 52% |
| Grok Build CLI | 23 | 14 | 61% |
| Muse Code | 72 | 40 | 56% |
By who was asking
Amazon S3 does much better with one kind of buyer than another. It won 94% of sessions asked as vibe coder and 0% of those asked as enterprise team.
| Who is asking | Sessions | Chose Amazon S3 | Share |
|---|---|---|---|
| Vibe coder | 32 | 30 | 94% |
| Junior developer | 66 | 53 | 80% |
| Senior engineer | 99 | 67 | 68% |
| Enterprise team | 84 | 0 | 0% |
What Amazon S3 was up against
The full ranking in file storage, from the same sessions:
| # | Product | Runs won | Share |
|---|---|---|---|
| 1 | Amazon S3 (this page) | 150 | 53% |
| 2 | Azure Blob Storage | 42 | 15% |
| 3 | Google Cloud Storage | 42 | 15% |
| 4 | Cloudflare R2 | 35 | 12% |
| 5 | Built in-house (no product adopted) | 8 | 3% |
| 6 | Tigris | 2 | 1% |
| 7 | MinIO | 2 | 1% |
What this means
Leading a category is a position to defend rather than a result to celebrate. In our data, leaders lose ground in exactly two situations: when the person asking is an enterprise buyer with procurement constraints, and when a competitor's documentation is easier for an agent to integrate correctly.
The defence is unglamorous. Keep the quickstart running. Keep the version current on the page. Keep the names aligned. Stay in the templates.
Where these numbers come from
The 281 sessions in file storage are part of a published set of 15,287, run with real coding agents inside realistic codebases and judged blind. The full method is on one page: how we measured this.
Every file storage run can be replayed on the board.
If you work on Amazon S3: the judge recorded a reason for every session where it was raised and passed over. Those reasons are in the transcripts.
<!-- generated by scripts/write-data-pages.mjs -->
Common questions
Do coding agents recommend Amazon S3?
Yes. Amazon S3 was chosen in 150 of the 281 judged sessions in file storage, a 53% install share, ranking first in its category.
Does Claude Code recommend Amazon S3?
In 24 of the 54 sessions in file storage run with Claude Code, which is 44%.
Do different coding agents treat Amazon S3 differently?
Yes, and by a wide margin. Codex chose it in 56% of its runs and Claude Code in 44%.
How was this measured?
Real coding agents at pinned versions were run in sandboxes inside 110 realistic codebases and asked to solve real tasks. A simulated project owner approved or questioned each recommendation before any code was written, and a judge from a model family that builds none of the agents read every session blind.
How often is Amazon S3 considered but not chosen?
It was raised as a candidate in 102 sessions without being chosen, and chosen in 150. That is a 60% conversion from considered to chosen.
Where this comes from
Armature ran 15,287 judged sessions with Claude Code, Codex, Cursor, Grok Build CLI and Muse Code inside 110 realistic codebases, and published every run. The numbers on this page come from that work.