🔍 Read the full analysis: AI Agents Now Start 4 In 10 GitHub Pull Requests—Review Is The Bottleneck on ThorstenMeyerAI.com
Get movie nights 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 started with a detectable coding agent in September 2026, up from 17% in March. The company introduced review and model-routing tools as agent-generated proposals grow, but a merge-rate increase does not establish that code is correct or safe.
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 GitHub’s Octoverse report. The company says that represents 110 million agent-started pull requests in one month, a surge that puts pressure on developers to review proposed changes even as automated tools make them faster to produce.
GitHub defines an agent-started pull request as one whose first draft came from a coding agent the company can detect. The measure includes public and private repositories, according to the report’s methodology, but GitHub did not provide a split between them. It says the count is a floor because some agent use may go undetected. A pull request can still count as agent-started if a human later substantially edits the draft.
GitHub reported that 90% of agent-started pull requests were merged within two weeks in September, compared with 83% in February. The merge rate for pull requests without a detected agent remained around 86%. Those figures show how often proposals were accepted, not whether the resulting code was correct, secure or free of defects. Agent-started pull requests also more often touched multiple files: 87% changed more than one file, against 79% of other pull requests.
The company announced several products focused on review and routing. Copilot code review Max combines reviews from multiple AI models into one result; Copilot approvals can count toward required approvals; and Copilot Fusion routes tasks to models it selects for each step. GitHub estimates an average review costs $0.25 in Lite mode, $1.50 in Balanced mode and $10 in Max mode, 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 Capacity Becomes a Constraint
The figures point to a change in where software teams may spend time: if agents produce more first drafts, review capacity can become the limiting step. Developers still need to judge whether a proposed change meets requirements, fits the surrounding system and introduces risks. Larger, multi-file changes can make that work more demanding, even when an agent has produced the initial code quickly.
For employers, the report raises practical questions about reviewer staffing, the cost of automated review and when an AI-generated approval should count toward a release requirement. Teams may need to track rework, reversions and incidents, not just how many pull requests agents open or how quickly they are merged. A higher merge rate alone cannot show that productivity or software quality improved.
For developers, the shift may increase the value of writing clear task instructions and critically reading code they did not write. It may also reduce the availability of small, routine coding tasks that have traditionally helped newer engineers learn. The report describes a platform-wide trend, however, and does not establish how broadly those job effects are occurring.
Popular right now
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
Vitamix Propel Series 510 Professional-Grade Blender, 3 Programs, Black
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent-Started
The figures come from GitHub’s annual Octoverse 2026 report, released October 28 at the company’s Universe conference. The report covers October 1, 2025, through September 30, 2026. GitHub’s wording matters: its headline measure counts a pull request when the first draft came from an agent it can detect, rather than measuring every contribution an agent may have made during development.
GitHub recalculated its 2024 and 2025 figures using updated definitions and a revised reporting window, so figures in older Octoverse reports are not directly comparable. The report also describes substantial overall platform activity: it records 1.3 billion merged pull requests during the year, compared with 538 million in 2025 and 402 million in 2024 under its recalculated figures. That wider growth is relevant to the review question, but the report does not attribute all of it to coding agents.
GitHub also says 55% of its monthly active Copilot users use agent features, up from 35% a year earlier, and reports that use of its cloud agent grew 230% year over year. These are company-reported platform metrics. They describe usage, not independent measures of code quality or the time saved by teams.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in its Octoverse report
What the Report Cannot Measure
GitHub has not published how the September figure divides between public and private repositories, or how many agent uses its detection misses. It also uses several related labels in the report, including “agent-started,” “agent-generated” and “agent-authored”; the precise definition for the headline figure is agent-started. The data does not show how much human rewriting took place after the first draft.
The report provides no evidence that the higher merge rate means agent-written code is more reliable. It does not provide comparable figures for defects, security problems, reversions or review time. GitHub says its cloud agent cannot approve or merge its own pull requests, and that the person who requested a change cannot approve it, but the report does not establish how organizations will set or enforce review rules across their systems.
How Teams Set Review Rules
GitHub’s new review and routing features give teams tools to test different levels of automated checking, with stated prices that vary by review mode. The next practical test will be whether those tools reduce reviewer workload without weakening review quality. The company has not supplied outcome data that answers that question.
Teams adopting agent workflows will need to decide where AI review can count toward required approvals, how much to spend on review credits and Actions minutes, and which changes still require human sign-off. Future reporting on review duration, rework, defects and reversions would help show whether the growth in agent-started pull requests is translating into safer or more efficient software delivery.
Key Questions
What does GitHub mean by an agent-started pull request?
It is a pull request whose first draft came from a coding agent GitHub can detect. A person may later edit the draft and it can still count in the reported figure.
How many pull requests did agents start in September?
GitHub reported 110 million in September 2026, or more than 40% of pull requests on the platform that month. The company says the measure is a floor because it may not detect all agent use.
Does a high merge rate mean the agent code was good?
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 did not provide matching defect or security data.
How much does GitHub estimate an AI review costs?
GitHub estimates $0.25 for Lite, $1.50 for Balanced and $10 for Max per average Copilot review. GitHub Actions minutes are additional.
Who is responsible for code written by an agent?
GitHub says developers remain accountable for what ships. Its cloud agent cannot approve or merge its own pull request, and the person who requested the change cannot approve it either.
Source: ThorstenMeyerAI.com
