🔍 Read the full analysis: GitHub Pull Request Review Faces A New Pace As AI Agents Start 4 In 10 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 says more than 40% of pull requests on its platform in September 2026 began with a coding agent’s first draft, representing 110 million pull requests. The company also introduced tools aimed at reviewing and routing agent-written code, but its figures measure detected agent use and a merge does not establish that code is correct.
Coding agents started more than 40% of GitHub pull requests in September, up from 17% in March, according to the platform’s Octoverse report released October 28. GitHub counted 110 million agent-started pull requests in September, a surge that makes code review and approval a larger part of the development process even as agents take on more first-draft coding.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. It says the count is a floor because some agent use may go undetected. The metric includes public and private repositories unless otherwise specified, but GitHub did not publish a breakdown by repository type. A pull request can still count as agent-started if a person later rewrites part of the draft.
The report says 90% of detected agent pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without detected agent involvement held at about 86%. Those figures describe whether a change was merged, not whether it was correct, safe or free of later rework.
At its Universe conference, GitHub announced review-focused products alongside the report. Copilot code review Max brings together reviews from several AI model providers into one combined review, while Copilot Fusion routes tasks to models for different steps. GitHub also made Copilot approvals eligible to count toward required approvals. The company estimates an average review costs $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 Becomes the New Constraint
When agents produce more proposed changes, teams must still decide which changes deserve review, who reviews them and what evidence is enough to approve them. A higher merge rate can speed delivery, but it does not by itself show that software quality has improved. GitHub’s figures make review capacity and oversight practical concerns for engineering teams, rather than questions that can be answered by counting generated code.
The change also affects budgets and work routines. Organizations may need to account for review credits and Actions usage, alongside the staff time required to inspect larger changes. The report says 87% of agent-started pull requests changed more than one file, compared with 79% of pull requests without detected agent use. That difference may mean reviewers have more connected changes to understand, although the statistic alone does not establish how much harder or riskier those reviews are.
For developers, the source material points to a shift in emphasis: writing clear task instructions and critically reading diffs become more central as agents handle more first drafts. For managers, useful measures include rework, reversions and incidents, not just pull-request volume or merge speed. Those are suggested ways to evaluate the change, not outcome data established by this report.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard Tactile (Graphite)
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent-Started
The Octoverse report covers October 1, 2025, through September 30, 2026. GitHub’s wording is important: its headline measure is about pull requests whose first draft came from a detectable coding agent. The report also uses nearby labels such as “agent-generated” and “agent-authored,” but “agent-started” most precisely describes the stated definition. The company says not all agent activity is detectable, so the reported share may not capture every agent-assisted change.
GitHub recalculated its 2024 and 2025 figures using updated definitions and a changed measurement window, so the historical figures are not directly comparable with earlier Octoverse reports. The company also reports broader growth in platform activity: 1.3 billion pull requests were merged during the report period, compared with 538 million in 2025 and 402 million in 2024. Those totals provide scale, but they do not isolate how much of the overall growth was caused by agents.
GitHub says 55% of Copilot’s monthly active users used agent features, up from 35% a year earlier, and use of its cloud agent rose 230% year over year. These are company-reported usage measures. They indicate adoption within Copilot but do not establish how users’ productivity or code quality changed.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, describing its measurement method in the Octoverse report
What the Adoption Figures Cannot Show
GitHub has not disclosed how the 110 million September pull requests divide between public and private repositories, nor the share of agent use its detection misses. The company’s own caveat means the reported 40%-plus rate is not a complete count of AI involvement. Because a human may substantially edit an agent’s draft, the metric also does not say how much of the final code an agent wrote.
The merge figures do not establish code quality or downstream effects. The report does not provide comparable data on defects, security issues, reverts, review time or maintenance costs for agent and non-agent pull requests. It is also not clear from the supplied report details how often teams will use Copilot approvals to satisfy required approvals, or what safeguards organizations will set around that option.
The stated review prices are GitHub estimates; the source does not give a typical company’s monthly bill or the cost of reviewer labor. The available information also does not show whether the newly announced tools will reduce review workload in practice.
Teams Face Review Policy Decisions
GitHub has released the review and routing features described at its October 28 Universe conference. Teams adopting them will need to decide where AI-generated reviews can count toward approval, what human checks remain mandatory and how to monitor spending, including Actions minutes. The source material does not give a rollout schedule or adoption results for the newly announced tools.
The next useful evidence will be operational: whether reviews catch defects, how much time reviewers spend on agent-started changes, and whether rework, reversions or production incidents change. GitHub’s next reporting cycle may provide updated platform figures, but teams will need their own measures to judge local results. For now, the September numbers show a rapid increase in detected agent-started work—not proof that review has become faster or that merged code is better.
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 later and it still counts. GitHub says undetected agent use is missing, so the reported share is a floor.
How many GitHub pull requests did agents start?
GitHub says agents started 110 million pull requests in September 2026, more than 40% of the month’s pull requests, compared with 17% in March.
Does a high merge rate mean agent-written code is reliable?
No. GitHub reports that 90% of detected agent pull requests were merged within two weeks in September, but merging does not prove correctness or quality. The report does not provide comparative defect or incident rates.
What will 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 additional, and the figures are estimates rather than a typical monthly bill.
Who remains responsible for code that an agent helps produce?
GitHub says developers remain accountable for what ships. Its cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve it either, according to the source material.
Source: ThorstenMeyerAI.com
