🔍 Read the full analysis: AI Agents Start 40% Of GitHub Pull Requests, Making Review The Next Hurdle on ThorstenMeyerAI.com
Get hardware and tech essentials 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 detectable coding agent, compared with 17% in March. The company also announced tools for AI-assisted review and approval, while saying developers remain accountable for changes that ship.
More than 40% of pull requests on GitHub in September began with a coding agent’s first draft, up from 17% in March, according to the company’s Octoverse 2026 report. GitHub said that amounted to 110 million agent-started pull requests in one month, as it introduced new review tools and rules for AI-assisted approvals.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. It says its count is a floor because not all agent use is visible to its systems. The report’s figure includes public and private repositories; GitHub did not provide a breakdown by repository type. A pull request is a proposed code change that is reviewed before it is incorporated into a project.
GitHub reported that 90% of detected agent-started pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without a detected agent held at about 86%. Those figures describe whether changes were merged, not whether the code was correct, safe or free of later rework.
At its October 28 Universe conference, GitHub introduced Copilot code review Max, which draws on several AI reviewers from different model providers and combines their feedback. It also made Copilot approvals eligible to count toward required approvals and released Copilot Fusion, a system that routes work to models for different steps. GitHub estimated the average review cost at $0.25 for Lite, $1.50 for Balanced and $10 for Max, plus GitHub Actions minutes.
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 Becomes the New Constraint
The figures suggest that producing a proposed change is becoming less of a bottleneck for some GitHub users, while checking and approving code may demand more reviewer time. Agent-started pull requests were also more likely to touch multiple files: GitHub said 87% changed more than one file, compared with 79% of pull requests without a detected agent. That can increase the amount of context a reviewer must examine, though the report does not quantify review time or error rates.
For engineering teams, the shift has practical costs. AI review is not free under GitHub’s published estimates, and teams still need people to assess whether changes fit their systems and requirements. Measuring success by the number of pull requests or how quickly they merge could miss later rework, reversions or incidents. GitHub’s data does not establish that agent use caused the overall increase in code activity or that merged agent changes perform as well as human-written ones.
The change also affects developers’ day-to-day work. If agents take on more first drafts, people may spend more time specifying tasks, inspecting diffs and deciding whether a proposed change is sound. The report does not measure effects on hiring or training, so claims about how the shift will affect junior developers remain uncertain.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard Tactile (Graphite)
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How GitHub Counts Agent Work
The Octoverse report covers October 1, 2025, through September 30, 2026. GitHub says its annual platform metrics generally include public and private repositories unless a metric says otherwise. For the agent figure, the company’s definition centers on the first draft and detectable agent activity; a human can still revise the code afterward and the pull request remains in the count.
GitHub’s report uses several labels, including “agent-started,” “agent-generated” and “agent-authored.” The source’s glossary definition is narrower and more specific: it counts pull requests whose first draft came from a detectable coding agent. GitHub also recalculated its 2024 and 2025 figures using updated definitions and a revised reporting window, so comparisons with older Octoverse reports may not be like-for-like.
The rise is part of a broader increase in activity on the platform. GitHub reported 1.3 billion merged pull requests during the report period, up from 538 million in 2025 and 402 million in 2024. Those totals cover all pull requests, not only agent-started ones. GitHub also said 55% of its monthly active Copilot users used agent features, compared with 35% a year earlier; use of its cloud agent grew 230% year over year, according to the report.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in its Octoverse report
What the Pull Request Data Misses
The reported more-than-40% share is based on detectable agent activity, and GitHub says some use goes uncounted. The company has not published a public-versus-private repository split for the figure, nor has it provided enough detail in the source material to independently assess how detection works across different agents and workflows.
Merge rates are not a measure of correctness. The report does not say how often agent-started changes were later reverted, required substantial rework, introduced security problems or caused production incidents. It also does not compare the time reviewers spent on agent-started and other pull requests. The available numbers show adoption and merging, but not the quality or total cost of the resulting code.
GitHub says its cloud agent cannot approve or merge its own pull requests, and that the person who requested a change cannot approve it either. The company’s announcement permits Copilot approvals to count toward required approvals, but the supplied material does not specify how organizations will configure that option or what safeguards apply in every workflow. It is also unclear how much the new review tools will change reviewer workloads once widely used.
How Teams Set Review Rules
The immediate next step for engineering organizations is deciding where AI review fits into their existing approval processes. Teams will need to determine whether Copilot approvals can satisfy any required checks, which changes still need human sign-off and how to account for the listed review prices and additional GitHub Actions usage.
GitHub’s report offers platform-wide adoption and merge figures, but does not provide a follow-up measure of code quality. Further reporting on rework, reversions, incidents and review time would help show whether faster first drafts translate into reliable software. The company has not provided a date for such results in the supplied material.
For now, GitHub’s position is that people remain responsible for code that ships. The practical test for teams will be whether their review controls can handle a larger flow of proposed changes without treating an AI-generated draft or an approval signal as proof that the change is correct.
Key Questions
What does GitHub mean by an agent-started pull request?
GitHub counts a pull request when its first draft came from a coding agent the company can detect. A person may edit that draft afterward and it still qualifies under the definition.
Did agents start more than 40% of all pull requests in September?
GitHub reported that more than 40% of its September pull requests began with a detectable agent. The company says the figure is a floor because some agent activity may go undetected.
Does a high merge rate mean the agent-written code was correct?
No. GitHub said 90% of detected agent-started pull requests were merged within two weeks, but merging does not establish correctness or quality. The report does not provide rates for rework, reverts or incidents.
How much does GitHub estimate AI code review costs?
GitHub estimated an average review at $0.25 for Lite, $1.50 for Balanced and $10 for Max, plus GitHub Actions minutes. The source does not give an estimate of total team costs at different usage levels.
Who is responsible for code that an agent helps produce?
GitHub says developers remain accountable for what ships. It also says its cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve it either.
Source: ThorstenMeyerAI.com
