AWS Emulator, which AWS launched on 11 November and says is built and maintained by agents, lists five commit authors in its public repository, and none of them is a person’s name, by our read of the full git history: IssueFinderAgent, IssueFixerAgent, Scaffolder, ReleaseManagerAgent and TraceIdMigration.
The same history shows the agents’ backlog. By our count, 12,708 issue files had moved to the repository’s fixed folder by the newest commit on 10 November, while 6,544 sat in the open folder, up from 1,930 on 28 October.
Business Pill · Multi-Agent Systems
An evergreen Business Engineer explainer on when one agent is enough and when a team of agents works better. It teaches the general idea only and does not discuss AWS Emulator.
The key insight: As we read it, the repository shows a test-and-fix loop run by agents at a scale that leaves a measurable queue: 12,708 issue files fixed and 6,544 still open at the newest commit, by our count, with a median of 21 hours from filing to fix.
IssueFinderAgent authored 16,664 commits and IssueFixerAgent 12,981, by our count of the aws/aws-emulator repository. Scaffolder authored 29, starting with the “Initial emulator scaffold” on 21 October, TraceIdMigration 6 and ReleaseManagerAgent 4.
Of those commits, 8,311 are merges, nearly all by IssueFinderAgent, so 21,373 carry changes of their own. All but six carry the author email aws-emulator@amazon.com. AWS’s launch post describes the split: “One group of agents looks for gaps”, while “Another group picks up an issue, writes tests, implements the behavior, and reviews the change for correctness, performance, and security”.
Issue files in the public aws/aws-emulator repository: the fixed folder (things-that-now-work) against the open folder (things-that-do-not-work), counted at the last commit before 00:00 UTC each day from 22 October 2026; the last point is the newest commit, 10 November 2026 20:39 UTC. Counts are our own.
Open Gaps Grew Faster Than Fixes
The README says individual gaps are tracked in the repository’s issues directory: what does not work yet sits in one folder, and fixed items move to another. We counted both folders at the last commit before midnight UTC on each day since 22 October.
The fixed folder grew from 138 files on 22 October to 12,708, about 66% of all issue files, by our arithmetic. The open folder fell to 1,930 on 28 October, then rose to 6,544, more than three times that level. From 1 November, the fixed folder added 2,460 files and the open folder 2,588.
The fixer slowed as the backlog grew. By our count, IssueFixerAgent made 1,526 fix commits on 22 October and 276 on 10 November, while 1,021 new issue files carry a 9 November date. An example open issue, rated “HIGH”, describes the S3 emulator storing an object under the wrong key, “the reverse of AWS”.
How Fast a Gap Closes
Each fix commit names the issue file it closes, and each file name carries the time it was filed. Across 12,590 such commits, the median gap between filing and fix was 21 hours, by our arithmetic, and 53% closed within a day.
A quarter closed in under 6.8 hours and a quarter took more than 61 hours. Another 94 IssueFixerAgent commits retract abandoned issues as “WontFix” rather than fixing them.
Five Scored Releases
Five of the repository’s 12 releases attach a fidelity score file. By our count, the pass rate rose from 55.8% in v1.0.290, scored on the morning of 10 November, to 56.5% in v1.0.322, scored just after midnight on 11 November.
That is 469 more passing tests, while the suite itself grew by 303 tests over the same five releases. The repository had 977 stars when we read it at about 21:20 UTC on 11 November, up from 458 about two hours earlier. Our launch piece on AWS Emulator covers the product and the single-release pass rates by service.
The Structural Read
As we read the two folders, the gap-finding side has outrun the fixing side since late October: the open folder more than tripled from its 28 October level while fix commits per day fell from 1,526 on 22 October to 276 on 10 November, by our count.
As we read the score files, the published pass rate moves in small steps between releases, from 55.8% to 56.5% across five scored releases, while the test suite itself grows.
AWS News Blog, 11 November 2026
“Because the code is written by agents, an independent security agent reviews changes adversarially, both while the fix is being made and again in the release pipeline as a final check before anything ships.”
Three Implications
FOR DEVELOPERS The README points users to the fidelity scores published with each release, and the open folder lists the gaps that do not work yet, so a team can check its own services before relying on the emulator.
FOR AWS Its launch post says one group of agents looks for gaps and another picks up issues, writes tests and implements the behavior; the repository records that division in its commit authors.
FOR THE MEASUREMENT OF AGENT-BUILT CODE By our count, 94 commits retract abandoned issues as WontFix, so the fixed folder and the open folder together do not count every issue ever filed.
The framework argues that what a technology can do and what users can do with it “are not the same curve, and the gap between them” has a name: overhang.
As we read it, AWS Emulator puts a version of that gap on public record: the score files measure how far the emulator is from AWS, and the open folder lists what is left. The framework predates the emulator and does not discuss it.
What Is Not Established
Author names in git are labels set by whoever makes the commit, so the history does not show whether, or how, people direct, review or approve the agents’ work. The GitHub API dates the repository to 11 September, while the public history starts on 21 October.
We counted issue files, not their severity, and a file can describe one gap or several. Our pass rates count each test in a score file once. We did not contact AWS for this piece.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
AWS Emulator’s public history lists no commit under a person’s name, by our count. Its agents have moved 12,708 issue files to the fixed folder, with a median of 21 hours from filing to fix, while the open folder holds 6,544, more than three times its 28 October level.
A note on sourcing. We cloned the public aws/aws-emulator repository on 11 November 2026 and counted its full git history ourselves: commit authors, merge commits, the fix commits that name the issue file they close, and the issue files in its open and fixed folders at the last commit before midnight UTC each day. Pass rates are our tally of the score files attached to the five releases that carry one, counting each test once. Quotes come from AWS’s 11 November launch post and the repository README. We did not contact AWS for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
