🔍 Read the full analysis: AI Agents Are Changing GitHub Pull Requests From Coding Through Review on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
GitHub reported that 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 introduced tools for AI-assisted review and model routing, but says people remain responsible for shipped code; the reported merge rate does not establish code quality.
More than 40% of GitHub pull requests in September 2026 began with a coding agent’s first draft, up from 17% in March, according to the company’s Octoverse report released October 28. GitHub says that amounts to 110 million agent-started pull requests in one month, a shift that is putting greater emphasis on reviewing and approving code rather than producing an initial draft.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. The company describes its count as a floor, because some agent use may not register in its data. A pull request still counts under that definition if a person later substantially rewrites the draft. The figure includes public and private repositories, but GitHub did not provide a breakdown between them.
The report says 90% of agent pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without a detected agent held at about 86%. A merge records that code was accepted into a project; it does not show whether the code was correct, required rework, or caused problems after release.
GitHub’s product announcements this week focus on review and workflow. Copilot code review Max brings together several AI reviewers from different model providers and returns a combined review. GitHub also made Copilot approvals eligible to count toward required approvals and released Copilot Fusion, a system that routes different steps to models. The company estimates average Copilot review costs of $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
When agents create more proposed changes, human attention can become a limiting resource. Teams still need people to assess whether a change fits the system, catches edge cases and is safe to merge. GitHub’s launches address that pressure by adding AI review and approval options, but those tools introduce their own questions about cost, reliability and where human sign-off remains necessary.
The reported merge rate may look like evidence that agent-generated code is readily accepted, but merging is not a quality measure. A team that tracks only the number of pull requests or merges could miss later rework, reversions and incidents. The practical test for companies is whether the new workflow improves delivery without weakening review or increasing downstream repair work.
For developers, the change raises the value of writing clear tasks and reading diffs critically. If agents take on small, bounded coding jobs, some work that once gave junior developers practice may become less available. The report does not establish how employment or training will change, but it points to a need for teams to make room for learning alongside automated production.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard Tactile (Graphite)
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
Logitech MX Keys S Wireless Keyboard, Graphite
As an affiliate, we earn on qualifying purchases.
How GitHub Counts Agent Work
The Octoverse 2026 report covers GitHub activity from October 1, 2025, through September 30, 2026. GitHub revised its 2024 and 2025 figures using updated definitions and a new reporting window, so older editions should not be compared directly with the latest figures without accounting for that change.
The report describes broad growth on the platform as well as agent use: it records 1.3 billion merged pull requests in the period, compared with 538 million in 2025 and 402 million in 2024. GitHub also said 55% of Copilot monthly active users now use agent features, up from 35% a year earlier. These are company-reported platform figures, not an independent assessment of code quality or productivity.
GitHub’s terminology varies across the report: some passages say “agent-started,” while charts use “agent-generated” or “agent-authored.” Its stated definition is narrower and more specific: the first draft must come from an agent GitHub can detect. That wording matters when interpreting or repeating the headline share.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in its Octoverse report
Limits of the Reported Numbers
GitHub has not published a public-versus-private repository split for the agent share, nor a measure of how many detected drafts were substantially rewritten by people before submission. Because the company says it cannot detect every agent, the actual share may differ from the reported figure, but the report does not quantify the gap.
The merge statistics do not reveal defect rates, review quality, later reverts or production incidents. The supplied report also does not establish whether AI reviews catch problems at a rate comparable to human reviewers, or how often their suggestions are accepted. Those outcomes will matter more than raw adoption if companies are to judge whether agent-assisted workflows improve software delivery.
Approval rules also leave implementation questions for individual teams. GitHub says the person who requested a change cannot approve it and that its cloud agent cannot approve or merge its own pull requests. The report does not spell out how organizations will configure AI approvals across different repositories or what safeguards they will use for sensitive code.
How Teams Measure the Shift
GitHub has announced the review and routing features, with usage priced according to review mode and additional GitHub Actions minutes. The next visible test will be how teams adopt them and whether AI-assisted review reduces reviewer workload without letting defects through. The source material does not provide a date for a broader outcome report or an independent evaluation.
Organizations adopting agent workflows will need to set approval policies, account for review credits and reviewer time, and compare results against their existing process. Measures such as rework, reversions and production incidents can show whether accepted changes hold up after merging. Developers and managers will also need to decide how to preserve opportunities for less experienced staff to learn through small coding tasks.
For now, GitHub’s report documents a rapid rise in detectable agent-started pull requests and a corresponding push toward AI-supported review. It does not establish that agents produce better code or that automated reviewers can replace human judgment. Those questions remain open as the tools enter routine use.
Key Questions
What share of GitHub pull requests started with an AI agent?
GitHub says more than 40% in September 2026 began with a detectable coding agent’s first draft, up from 17% in March. The company reported 110 million such pull requests in September.
Does a merged agent pull request mean the code was correct?
No. A merge means the change was accepted into a repository. GitHub’s reported merge rate does not measure defects, later rework, reversions or production quality.
How does GitHub define an agent-started pull request?
It is a pull request whose first draft came from a coding agent GitHub can detect. The count can include a draft later substantially rewritten by a person, and GitHub says some agent use may go undetected.
What did GitHub announce for AI code review?
GitHub introduced Copilot code review Max, which combines reviews from several AI reviewers; made Copilot approvals eligible to count toward required approvals; and released Copilot Fusion for routing models across workflow steps.
Who remains responsible for code written by an agent?
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 requesting a change cannot approve it.
Source: ThorstenMeyerAI.com
