🔍 Read the full analysis: GitHub Teams Face A Review Shift As AI Agents Start 4 In 10 Pull Requests on ThorstenMeyerAI.com
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
GitHub’s Octoverse report says more than 40% of pull requests on its platform in September 2026 began with a coding agent’s first draft, up from 17% in March. The company also announced AI review and model-routing tools as it shifts attention toward checking and approving the growing volume of agent-written code.
More than 40% of pull requests on GitHub started with a coding agent’s first draft in September, up from 17% in March, according to the company’s 2026 Octoverse report. The figures point to a change in the work facing software teams: as agents write more proposed code, developers must spend more time reviewing it, and GitHub has introduced new tools for that process.
GitHub said the September figure represents 110 million agent-started pull requests in one month, about six times March’s volume. It defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. GitHub describes the figure as a floor because its systems may miss some agent use. The report includes public and private repositories unless a metric is specifically marked otherwise; GitHub did not provide a public-versus-private breakdown for this figure.
The report also says 90% of agent pull requests were merged within two weeks in September, compared with 83% in February. The reported rate for pull requests without a detected agent remained about 86%. A merge means code was accepted into a repository; it does not establish that the change was correct, safe or free of later rework.
At its October 28 Universe conference, GitHub introduced Copilot code review Max, which brings together reviews from multiple model providers into one combined review, and Copilot Fusion, which routes different steps to models. The company also said Copilot review can count toward required approvals. Its estimated average review prices are $0.25 for Lite, $1.50 for Balanced and $10 for Max, with GitHub Actions minutes charged separately.
Infographic 01 · The agent share
Four in ten pull requests now start with an agent.
GitHub’s own telemetry, March to September 2026. The first draft comes from an agent; review and merge stay with people.
Merged within two weeks
Share of pull requests merged within 14 days
Merged fast is not the same as correct. GitHub publishes no revert or defect rate.
Bigger changes to review
Share of pull requests that change more than one file
More files per change means more for a reviewer to check.
What GitHub counts: an “agent-started pull request” is one “whose first draft came from a coding agent that we can detect.” Undetected agents are missing, so GitHub calls the figure a floor. Metrics cover public and private repositories unless noted.
Source: GitHub Octoverse 2026, 28 Oct 2026 (text, chart captions, glossary and methodology). All figures are GitHub’s own.
Powered by Thorsten Meyer AI
Review Work Becomes the Bottleneck
The figures suggest that software teams may face a different capacity problem as coding agents become more common. If agents produce more initial drafts, the limiting work can shift from writing code to checking changes, tracing their effects and deciding whether to approve them. The report says 87% of agent pull requests changed more than one file, compared with 79% of pull requests without a detected agent, potentially giving reviewers a broader set of changes to inspect.
That shift has practical consequences for engineering managers and developers. Teams may need to plan for reviewer time and review-tool costs alongside computing costs, while tracking outcomes such as rework, reversions and incidents rather than treating the number of pull requests as a measure of success. GitHub’s merge statistics show how often changes entered repositories, not whether those changes performed well after release.
For individual developers, the report’s trend puts greater emphasis on writing clear task instructions and reading diffs critically. Agents can help produce a first draft, but a human still has to judge whether it fits the intended behavior and whether the change is ready to ship. That responsibility is especially relevant for junior developers, whose early work has often included smaller tasks that agents may now handle.
Products Worth Considering
Therabody Theragun Mini 3rd Generation Massage Gun, Black
Levoit Eleva 300X HEPA Air Purifier for Home and Bedroom
Philips Sonicare DiamondClean Smart 9300 Electric Toothbrush, Black
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent Work
The Octoverse is GitHub’s annual report on activity across its platform. The 2026 edition covers October 1, 2025, through September 30, 2026. GitHub’s definition counts a pull request when the first draft came from an agent it can detect; it does not mean an agent wrote every line that was eventually submitted. A person may substantially revise the draft before opening or completing the request.
The report uses several related descriptions, including “agent-started,” “agent-generated” and “agent-authored.” Its stated definition is specifically about the first draft, so “agent-started” is the most precise description of the headline metric. GitHub also recalculated 2024 and 2025 figures using updated definitions and the report’s time window, meaning those figures are not directly comparable with older Octoverse editions.
The broader platform figures show the scale of activity during the same period: GitHub reported 17.9 billion code pushes and 34.3 billion commits in the 12 months to September. Copilot’s monthly active users who used agent features rose to 55%, from 35% a year earlier, while use of the Copilot cloud agent grew 230% year over year, according to GitHub. These are company-reported platform measures, not an independent assessment of code quality or productivity.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in its definition of an agent-started pull request
What the Figures Cannot Show
GitHub has not published a breakdown showing how the agent-started share differs between public and private repositories, or by company, team size or programming language. Because the metric includes only agents GitHub can detect, the actual share may be higher, but the company has not quantified how much higher. The report’s different labels for agent-related pull requests also make its explicit first-draft definition important when comparing or quoting the figures.
The merge rate does not show whether a pull request introduced defects, needed later correction or improved a product. GitHub’s published statistics do not establish that agent use caused changes in overall code quality or team productivity. The source material also does not specify how organizations will configure Copilot review as a required approval in different settings, or how they should assess the reliability of a combined review from multiple models.
Teams Weigh Review Costs
GitHub has announced the review and routing products, with estimated prices for its three Copilot review levels. Teams adopting them will need to decide where AI review fits into their approval rules, how many human reviews remain necessary and how to account for review charges and GitHub Actions minutes. The source material does not provide a timetable for broader adoption or independent results comparing the tools with human review.
GitHub says its cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve it either. Those rules preserve a human approval step, although the degree of human scrutiny in practice will depend on each team’s process. The next useful evidence will be whether companies report changes in rework, reversions, incidents and review time as agent-generated drafts become more common.
Key Questions
What does GitHub mean by an agent-started pull request?
It means the pull request’s first draft came from a coding agent GitHub can detect. A person may revise the draft, and the metric does not mean every line was written by an agent.
How many pull requests did agents start in September?
GitHub reported 110 million agent-started pull requests in September 2026, representing more than 40% of pull requests on the platform under its definition.
Does a high merge rate prove that agent-written code is good?
No. GitHub reported that 90% of agent pull requests were merged within two weeks, but merging is not a measure of correctness or long-term quality. The figures do not report defects, reversions or later rework.
How much does Copilot code review cost?
GitHub estimates an average review at $0.25 for Lite, $1.50 for Balanced and $10 for Max. GitHub Actions minutes are an additional cost.
Who is responsible for code that an agent helps write?
GitHub says developers remain accountable for what ships. The company says its cloud agent cannot approve or merge its own pull request, and the person who requested the change cannot approve it either.
Source: ThorstenMeyerAI.com
