🔍 Read the full analysis: GitHub’s Pull Request Workflow Is Changing As AI Agents Start 4 In 10 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 2026 Octoverse report says more than 40% of pull requests on its platform in September began with a coding agent’s first draft, up from 17% in March. The company also announced AI review and routing tools as review workloads grow, but says developers remain accountable for code that ships. Its count covers agents GitHub can detect, and a merged pull request is not proof that the code is correct.
More than four in 10 pull requests on GitHub in September began with a coding agent’s first draft, up from 17% in March, according to the company’s 2026 Octoverse report, released October 28 at its Universe conference. GitHub said the September total was 110 million agent-started pull requests, as it introduced new tools for reviewing and routing code written with AI agents.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. It calls the resulting share a floor because some agent use may not be captured by its systems. The report’s methodology says figures include public and private repositories unless a metric is specifically excluded; GitHub did not provide a public-versus-private breakdown for the agent figure.
The company said 90% of agent-started pull requests were merged within two weeks in September, compared with 83% in February. Pull requests without a detected agent had a merge rate of about 86% over the reported period. Merging indicates that a change was accepted into a codebase; it does not establish that the code was correct, safe or free of later rework.
GitHub’s conference launches focused on review rather than generating more code. Copilot code review Max uses several AI reviewers from different model providers and combines their feedback. GitHub also made Copilot approvals eligible to count toward required approvals and released Copilot Fusion, which routes different tasks to models it selects. The cloud agent cannot approve or merge its own pull requests, and the person who requested a change cannot approve it, GitHub said.
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 Capacity Becomes the Constraint
The numbers point to a change in where development work may accumulate: if agents produce more first drafts, teams need enough people and review systems to evaluate those drafts. GitHub’s product announcements reflect that pressure, placing review, routing and approval alongside code generation as parts of an AI-assisted workflow.
There is also a budgeting consequence. GitHub estimates an average Copilot review costs $0.25 for Lite, $1.50 for Balanced and $10 for Max, plus GitHub Actions minutes. Those estimates make automated review a metered expense, not a cost-free substitute for human attention. Companies will need to decide which changes merit which review level and how much reviewer time to set aside.
For developers, a higher volume of proposed changes can make clear task definitions and careful reading of diffs more important. GitHub says developers remain accountable for what ships. Teams assessing whether agents help should look beyond pull-request volume or merge rates to measures such as rework, reversions and production incidents.
Products Worth Considering
Sonos Era 100 SL Compact, Microphone-Free Speaker – Black
Ring Battery Doorbell (2nd Gen) and Indoor Cam Plus Bundle, Matte Mocha and Black
Dyson V8 Cyclone Cordless Vacuum
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent Activity
The report covers October 1, 2025, through September 30, 2026, and presents a year of broad growth in platform activity. GitHub reported 17.9 billion code pushes and 34.3 billion commits during the period. Merged pull requests rose from 402 million in 2024 and 538 million in 2025 to 1.3 billion in the latest report. These totals provide scale for the growth in agent activity, but they do not show that agents alone caused the increase.
GitHub said agent-started pull requests grew to roughly six times their March volume by September. Its report also says 87% of agent-started changes touched more than one file, compared with 79% of pull requests without a detected agent. That difference could mean reviewers have more files to assess in many agent-started changes, though file count alone does not measure complexity or quality.
The report’s terminology varies: it uses “agent-started,” “agent-generated,” “agent-authored” and “agent-opened” in different places. The defined measure is narrower and more precise: a pull request whose first draft came from a detectable coding agent. GitHub also recalculated 2024 and 2025 figures using its updated definitions and reporting window, so the revised historical data should not be treated as directly comparable with older Octoverse reports.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, describing its agent-started pull request measure
Limits of the September Count
GitHub has not published the share of agent-started pull requests in public versus private repositories, nor a breakdown by company, project size or type of agent. Because its systems detect only some agent activity, the reported 40%-plus share is not a complete count of every pull request that involved an AI tool. GitHub’s use of several labels across the report also makes the defined “first draft” measure important when comparing or quoting the figure.
The merge data does not answer whether agent-started changes have higher error rates, require more follow-up work or cause more incidents after release. The report gives no direct quality comparison that would settle those questions. It is also unclear how often teams will use each Copilot review tier or how the new AI approval option will affect human review practices in day-to-day work.
How Teams Put Reviews to Work
GitHub has released the review and routing products described at its October 28 conference, but adoption and outcomes will take time to assess. The company’s next Octoverse report or other product data may show whether agent-started pull requests continue to grow and how their review and merge patterns change.
For engineering teams, the immediate decisions are practical: choose when Copilot approvals can count, set review budgets, and determine which changes need human approval or more intensive checks. Tracking rework, reversions and production incidents alongside merge rates would help distinguish faster throughput from better software. GitHub’s current figures establish a rapid increase in detectable agent-drafted changes; they do not yet establish their longer-term effect on code quality.
Key Questions
What does GitHub mean by an agent-started pull request?
GitHub defines it as a pull request whose first draft came from a coding agent it can detect. A person may still edit the draft before review or merge.
Did AI agents start exactly 40% of GitHub pull requests?
No. GitHub said the share was more than 40% in September 2026. It describes the figure as a floor because its systems may miss some agent use.
Does a high merge rate show that agent-written code is good?
No. GitHub reported that 90% of agent-started pull requests were merged within two weeks in September, but merging is not a measure of correctness or safety. The report does not establish comparative post-release quality.
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, plus GitHub Actions minutes. Actual spending will depend on use.
Who is responsible for code an agent helps write?
GitHub says developers remain accountable for code that ships. Its cloud agent cannot approve or merge its own pull request, and the person who requested the change cannot approve it.
Source: ThorstenMeyerAI.com
