🔍 Read the full analysis: When AI Agents Start GitHub PRs, Review Becomes A Bigger Part Of The Work on ThorstenMeyerAI.com
Get the latest gadgets 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 detectable coding agents started more than 40% of pull requests in September 2026, up from 17% in March. The company introduced new review and model-routing tools as agent-written changes become more common, while saying developers remain accountable for code that ships.
Coding agents started more than 40% of pull requests on GitHub in September 2026, up from 17% in March, according to the company’s new Octoverse report. GitHub says that amounted to 110 million agent-started pull requests in September, a rise that is shifting attention from producing code to checking and approving it.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the platform can detect. The company says its count is a floor because some agent use may go undetected. The figure includes public and private repositories, according to the report, but GitHub did not provide a split between them. A pull request is a proposed change that is reviewed before it is incorporated into a codebase.
The report says 90% of detected agent pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without a detected agent remained at about 86%. Those figures describe merging, not whether a change was correct, safe, or free of later rework. GitHub also reports that 87% of agent-started pull requests changed more than one file, compared with 79% of pull requests without a detected agent.
Alongside the report, GitHub announced review and routing products. Copilot code review Max brings together reviews from several model providers into one combined review; Copilot approvals can count toward required approvals; and Copilot Fusion routes tasks to models for different steps. GitHub estimates a Copilot review costs $0.25 for Lite, $1.50 for Balanced and $10 for Max, plus GitHub Actions minutes. The source material does not give more detail about the terms or availability of each feature.
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 Is Becoming the Constraint
The figures suggest that generating a first draft is becoming a smaller share of the work around a code change. When agents produce more proposed changes, teams must still check what the code does, judge its fit and decide whether it should ship. A high merge rate alone cannot answer whether those changes perform well after release.
The new review options also put a direct price on part of that workflow. Organizations adopting agents may need to budget not just for code-generation tools, but for review usage, reviewer time and additional automation costs. GitHub’s per-review estimates offer a starting point, but the source does not establish how costs will vary across projects or how much human review the tools can replace.
For developers, the change increases the value of writing clear tasks and examining diffs critically. The source argues that routine starter tasks, often used by junior developers to learn, are among the work agents can take on. How that affects training and entry-level roles is not established by the report, but it gives teams a reason to think about how new developers will gain practical experience.
Popular right now
De'Longhi Rivelia Automatic Espresso Machine with Milk Frother
KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, Almond Cream
Roborock F25 Ultra Steam Wet Dry Vacuum Mop
As an affiliate, we earn on qualifying purchases.
How GitHub Counts Agent Drafts
The Octoverse is GitHub’s annual review of activity on its platform. The 2026 edition covers October 1, 2025, through September 30, 2026. GitHub recalculated figures for 2024 and 2025 using revised definitions and a revised measurement window, so comparisons with older Octoverse reports may not be direct.
The report uses several descriptions for the activity, including “agent-started,” “agent-generated” and “agent-authored.” Its stated definition focuses on whether the first draft came from a detectable coding agent. That does not mean every line in the final pull request was written by an agent: a human may have changed substantial portions afterwards. The published count also cannot show how much agent use went undetected.
The report places the growth alongside a broader rise in GitHub activity: it records 1.3 billion merged pull requests during the year, compared with 538 million in 2025 and 402 million in 2024. GitHub also says 55% of Copilot monthly active users now use agent features, up from 35% a year earlier. These are platform figures reported by GitHub, not an independent measure of coding-agent use across the software industry.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in the Octoverse report
What the Adoption Figures Cannot Show
The report does not say what share of agent-started pull requests came from public repositories versus private company code, or how many agent-assisted changes its detection misses. Because the measure follows the first draft, it also does not quantify how much humans revised each proposal before submission or merge.
The merge figures do not establish code quality or downstream outcomes such as incidents, reversions, or maintenance costs. GitHub’s reported rise in agent activity is platform telemetry; the material provides no independent verification or comparable industry-wide estimate. It is also unclear how much review teams can delegate safely to Copilot, how its estimated prices translate to typical monthly bills, and whether the new review products will reduce human review time.
Teams Must Track Outcomes After Merge
The next test for companies adopting these tools is whether faster drafting and new review systems improve delivery without increasing defects or follow-up work. Teams can track rework, reversions, incidents and review time alongside pull-request volume and merge rates. The source material does not identify a scheduled follow-up report or a deadline for broader product availability.
GitHub’s stated safeguards leave people responsible for sign-off: its cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve it either. Teams deciding whether Copilot approvals count toward required approvals will still need to set their own review policies. Further details about adoption, costs and real-world quality remain to be established.
Key Questions
How many GitHub pull requests did agents start in September?
GitHub says detectable coding agents started 110 million pull requests in September 2026, representing more than 40% of pull requests on the platform that month.
What does GitHub mean by an agent-started pull request?
It means the first draft came from a coding agent GitHub could detect. The count can miss undetected agent use, and it does not mean the final proposal was left unchanged by a human.
Does a high merge rate prove that agent-written code is good?
No. GitHub reports that 90% of detected agent pull requests were merged within two weeks in September, but merging is not evidence by itself that code was correct or performed well after release.
How much does a Copilot code review cost?
GitHub estimates an average review at $0.25 for Lite, $1.50 for Balanced and $10 for Max, plus GitHub Actions minutes. These are the company’s estimates; the material does not provide typical total monthly costs.
Who is responsible for code an AI agent produces?
GitHub says developers remain accountable for code that ships. Its cloud agent cannot approve or merge its own pull requests, and a person who requested a change cannot approve it either.
Source: ThorstenMeyerAI.com
