🔍 Read the full analysis: AI Agents Are Changing Where GitHub Teams Spend Pull Request Time 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’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 tools for AI-assisted review and model routing, while keeping approval and responsibility with human developers. The reported merge rate does not establish that agent-written changes were correct or safe.
More than 40% of pull requests on GitHub in September began with a coding agent’s first draft, compared with 17% in March, the company said in its Octoverse 2026 report, released October 28. The shift is changing where development teams spend time: with agents producing more initial code, GitHub is adding tools for review, routing and approval, while saying developers remain responsible for what gets shipped.
GitHub counted 110 million agent-started pull requests in September. It defines the measure as a pull request whose first draft came from a coding agent the company can detect. GitHub describes the count as a floor because some agent use may not be detected. The figure includes public and private repositories, according to the report, but GitHub did not provide a split between them.
The report says 90% of agent pull requests were merged within two weeks in September, up from 83% in February. Pull requests without a detected agent held at about 86%. A merge records that a change was accepted into a codebase; it does not, by itself, show that the code was correct, secure or free of later rework.
GitHub’s new products focus on review and workflow controls. Copilot code review Max gathers reviews from several AI model providers and combines them. GitHub also made Copilot approvals eligible to count toward required approvals and introduced Copilot Fusion, a router that selects models for different steps. The company estimates an average review costs $0.25 for Lite, $1.50 for Balanced and $10 for Max, with GitHub Actions minutes billed 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 Work Is Becoming the Constraint
The reported change suggests that teams may need to shift time and resources from producing first drafts to checking, testing and approving them. Agent-written changes are not necessarily small: GitHub says 87% of agent pull requests changed more than one file, compared with 79% of pull requests without a detected agent. A change spanning more files can give reviewers more interactions and dependencies to assess, though the report does not measure how long reviews take or how often they find defects.
For engineering leaders, the practical question is not simply how many pull requests agents produce. It is whether teams can review that output without adding unacceptable costs or risk. The per-review estimates make AI review a budget item, alongside the human time and Actions minutes involved. Teams will need to judge the tools against outcomes such as rework, reverted changes and production incidents, rather than treating a rising merge count as proof of better software.
The shift also affects developers’ work. Writing a clear task for an agent and carefully reading its changes become more consequential when the initial draft is automated. The report’s adoption figures do not establish how jobs or entry-level roles will change, but they make it relevant for teams to examine how junior developers gain experience when routine starter tasks can be delegated to agents.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard Tactile (Graphite)
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How GitHub Defines Agent Activity
The figures come from GitHub’s annual Octoverse report, which covers October 1, 2025, through September 30, 2026. The report says developers pushed 17.9 billion times during that period and merged 1.3 billion pull requests, compared with 402 million in 2024 and 538 million in 2025. GitHub recalculated its 2024 and 2025 figures using updated definitions and a revised window, so those historical totals are not directly comparable with earlier Octoverse reports.
The wording matters when describing the headline. GitHub’s definition is “agent-started”: a detectable agent supplied the first draft. A person may have substantially edited the change afterward and it still qualifies. The report also uses terms such as “agent-generated,” “agent-authored” and “agent-opened” in other places. Those phrases are not necessarily interchangeable with the defined measure, so the 40%-plus figure should not be read as the share of code written entirely by agents.
GitHub also reported that 55% of Copilot’s monthly active users used agent features, up from 35% a year earlier, and that use of its cloud agent grew 230% year over year. These are company-reported product metrics, not independent measures of code quality. The report says agent-related activity surged around June, when GitHub described agents as moving from single-session tools toward persistent cloud-based work that can continue across days.
“A pull request whose first draft came from a coding agent that we can detect.”
— GitHub, in its Octoverse 2026 report
What the Adoption Figures Cannot Show
The 40%-plus figure is GitHub telemetry, not an independently audited estimate, and the company says its detection misses some agent use. It has not disclosed how much activity goes undetected or provided a public-versus-private repository breakdown. The report’s varying labels also leave room for confusion when the headline is repeated: the stated definition is based on who supplied the first draft, not on how much code an agent ultimately contributed.
Nor do the published numbers answer whether agent-started pull requests produce better or worse code. The reported two-week merge rate does not measure correctness, security defects, maintenance burden, review effort, later reversions or incidents. GitHub’s report provides adoption and platform activity measures, but the supplied material does not provide a comparable quality study or independent verification of the customer outcomes mentioned elsewhere in the report.
It is also not clear how widely teams will use AI-generated approvals to satisfy repository rules, or how companies will set acceptable review budgets. GitHub’s stated controls describe its cloud agent’s approval limits, but the report does not detail how every organization configures its own review policies or how responsibility is assigned across teams in practice.
Teams Must Measure Review Outcomes
GitHub’s announced review products give teams new options for combining AI checks with existing pull request workflows. The immediate next step for organizations is to decide where automated review can count, what changes still require human approval and how review costs fit their budgets. The source material does not specify a future date for broader rollout or report adoption rates for the new features.
As agent use expands, engineering teams will need to compare the time saved drafting code with time spent examining it. Useful measures include review turnaround, defects caught before release, subsequent rework and reversions, and production incidents. GitHub’s headline numbers establish that detectable agent-started pull requests have become a substantial share of platform activity; whether that translates into lasting gains depends on evidence about outcomes that the report does not yet provide.
Key Questions
What does GitHub mean by an agent-started pull request?
It means GitHub detected that a coding agent supplied the pull request’s first draft. A developer may edit the draft afterward and it still counts.
How many GitHub pull requests started with agents in September?
GitHub reported 110 million, representing more than 40% of pull requests on the platform that month. It says the total is a floor because some agent activity may not be detected.
Does the merge rate prove agent-written code is reliable?
No. GitHub reported that 90% of agent pull requests were merged within two weeks, but a merge is not a measure of correctness, security or future maintenance costs.
What does AI code review cost under GitHub’s estimates?
GitHub estimates an average review at $0.25 for Lite, $1.50 for Balanced and $10 for Max. GitHub Actions minutes are additional.
Who is responsible for code an agent submits?
GitHub says developers remain accountable for shipped code. It 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
