Data study · The open-source tooling around AI search
Pull Requests Adding llms.txt Files Grew 3.3 Times as a Share of GitHub Activity Between Late 2025 and Late 2026
We counted 721 pull requests opened from llms.txt branches between October 2025 and October 2026. Their rate rose from 6.3 per million recorded pull request events in Q4 2025 to 20.6 per million in Q3 2026.
Analysis of 721 llms.txt pull requests in 622 repositories, Oct 2025 to Oct 2026 · GH Archive via BigQuery, October 2026
Key findings
- 1
Developers opened 721 pull requests from llms.txt branches in 622 public repositories between October 2025 and October 20, 2026.
- 2
The rate rose from 6.3 per million recorded pull request events in Q4 2025 to 12.1 in Q1 2026, 16.8 in Q2 and 20.6 in Q3, a 3.3-fold increase.
- 3
Merges kept pace: 536 merge events were recorded from December 2025, against 627 opened events in the same months (85.5%).
- 4
Where both opening and merging were recorded, the median llms.txt pull request was merged in 9 minutes. 70 of 105 (66.7%) were merged within an hour.
- 5
Documentation and website repos account for 158 of the 721 pull requests (21.9%), and their merge events matched 83.5% of their openings.
- 6
Adding a new file is the most common branch intent: 259 (35.9%) are named for adding llms.txt, 123 (17.1%) for fixing or updating one and 65 (9.0%) for generating it automatically.
Summary
Picture a software company's marketing lead who asks the developer team to 'add that AI file everyone talks about' to the docs site. The developer opens a branch called add-llms-txt, and the change is live within the hour. That small task is now happening in hundreds of projects.
llms.txt is a plain-text file that points AI tools to a site's most useful pages. Documentation teams were early adopters because AI coding assistants read docs constantly. The change usually arrives as a pull request on GitHub.
We counted 721 pull requests from llms.txt branches over the past year. Measured against all GitHub pull request activity, their rate grew 3.3 times between late 2025 and late 2026. Most get merged quickly.
What we measured
We used GH Archive, the public log of GitHub events, in Google BigQuery. Since October 2025, GH Archive's pull request records no longer include titles, but they still include the source branch name. We counted pull request events whose branch name contains llms.txt, llms-txt, llmstxt or llms-full.txt.
We counted opened and merged events by month and divided by all pull request events GH Archive recorded that month, because its capture varied a lot during 2026. Merge events appear in the data from December 2025. We paired opened and merged events for the same pull request to measure time to merge, and grouped repos by whether their name marks them as docs or website repos.
llms.txt pull requests keep gaining ground
Pull requests opened from llms.txt branches each month, and the rate per million recorded pull request events.
Pull requests
| Month | Opened | Merged | All recorded PR events | Opened per million PR events |
|---|---|---|---|---|
| 2025-10 | 51 | n/a | 6,268,230 | 8.1 |
| 2025-11 | 43 | n/a | 6,653,352 | 6.5 |
| 2025-12 | 45 | 36 | 9,184,361 | 4.9 |
| 2026-01 | 76 | 71 | 9,208,937 | 8.3 |
| 2026-02 | 114 | 98 | 8,352,302 | 13.6 |
| 2026-03 | 99 | 74 | 6,235,706 | 15.9 |
| 2026-04 | 77 | 66 | 5,378,552 | 14.3 |
| 2026-05 | 69 | 59 | 3,493,300 | 19.8 |
| 2026-06 | 23 | 17 | 1,163,246 | 19.8 |
| 2026-07 | 3 | 4 | 325,714 | 9.2 |
| 2026-08 | 7 | 6 | 348,205 | 20.1 |
| 2026-09 | 105 | 91 | 4,896,011 | 21.4 |
| 2026-10 (to 20th) | 9 | 14 | 943,646 | 9.5 |
Source: aisearchoptimization.net analysis of 721 llms.txt pull requests in 622 repositories, Oct 2025 to Oct 2026 · GH Archive via BigQuery, October 2026
Raw counts fall in June to August 2026 because GH Archive captured far fewer events then. The per-million rate shows steady growth: 6.3 in Q4 2025, 12.1 in Q1 2026, 16.8 in Q2 and 20.6 in Q3.
Most llms.txt changes are merged within minutes
For 105 pull requests opened between December 2025 and August 2026, GH Archive recorded both the opening and the merge.
Pull requests
| Merged within | Pull requests | Share of 105 (%) |
|---|---|---|
| 1 hour | 70 | 66.7 |
| 24 hours | 96 | 91.4 |
| 7 days | 102 | 97.1 |
Source: aisearchoptimization.net analysis of 721 llms.txt pull requests in 622 repositories, Oct 2025 to Oct 2026 · GH Archive via BigQuery, October 2026
Adding llms.txt is a low-risk change, and some teams generate it automatically, which may be why it rarely waits long for review. Pairs are only counted when GH Archive captured both events, so slow merges may be underrepresented.
Docs repos are a fifth of the activity
We split repos by name and grouped branch names by intent.
| Group | Opened | Share (%) |
|---|---|---|
| Docs or website repo | 158 | 21.9 |
| Code or other repo | 563 | 78.1 |
| Branch: add llms.txt | 259 | 35.9 |
| Branch: fix or update | 123 | 17.1 |
| Branch: generator or automation | 65 | 9 |
| Branch: llms-full.txt | 17 | 2.4 |
| Branch: other | 257 | 35.6 |
Source: aisearchoptimization.net analysis of 721 llms.txt pull requests in 622 repositories, Oct 2025 to Oct 2026 · GH Archive via BigQuery, October 2026
Many code repos host their documentation in the same repository, so the docs share is a floor. One in six llms.txt pull requests fixes or updates an existing file, a sign that teams maintain it after the first version.
What this means
Software and documentation teams are treating llms.txt as routine housekeeping. It takes minutes to add and rarely meets resistance. For a business website the same logic applies: it is cheap, low risk and easy to keep updated if your platform can generate it.
Do not expect it to carry your AI visibility on its own, though. AI assistants still lean on your actual page content and what other sites say about you. Both get checked in a free AI visibility audit, plus your llms.txt file, so you can see which parts of LLM SEO actually need work.
Limits
Since October 2025 GH Archive does not include pull request titles, so we rely on branch names. Pull requests that add llms.txt from a branch with a different name are missed, so the true count is higher.
GH Archive captured far fewer events in May to August 2026 and had gaps in October 2025 and August to October 2026. We use per-million rates to compare periods.
Merge events appear in the data from December 2025. Time to merge is measured only where both events were captured, and some fast merges are automated.
Repo type comes from repository names, so docs hosted inside code repos are counted as code repos. We could not identify the documentation framework from the event data.
Embed this research
Add the charts from this study to your page. The embed links back here.
<iframe src="https://aisearchoptimization.net/embed/data-studies/llms-txt-pull-requests-github" title="llms.txt Pull Requests on GitHub Tripled in Share in a Year" width="100%" height="720" style="border:0" loading="lazy"></iframe>
Cite this study
APA
Grenyo, K. (2026). llms.txt Pull Requests on GitHub Tripled in Share in a Year. AI Search Optimization. https://aisearchoptimization.net/resources/data-studies/llms-txt-pull-requests-github
Plain link
llms.txt Pull Requests on GitHub Tripled in Share in a Year - AI Search Optimization - https://aisearchoptimization.net/resources/data-studies/llms-txt-pull-requests-github
Frequently asked questions
How many projects added llms.txt on GitHub in the past year?
We counted 721 pull requests from llms.txt branches in 622 public repositories between October 2025 and October 20, 2026. That is a minimum because other branch names are not counted.
Is llms.txt adoption growing among developers?
Yes. llms.txt pull requests rose from 6.3 per million recorded pull request events in Q4 2025 to 20.6 per million in Q3 2026, a 3.3-fold increase.
Do teams actually merge llms.txt changes?
Mostly yes. Recorded merge events equaled 85.5% of opened events from December 2025. Where both were captured, the median merge came after 9 minutes.
Should my business website have an llms.txt file?
It is cheap and low risk, which is why developers merge it quickly. It is meant to help AI tools find key pages, but your page content and reputation matter more for AI answers.
Related studies
- The open-source tooling around AI searchGEO and AI SEO Repos on GitHub Hit 322 a Month in 2026
- The open-source tooling around AI search45% of GitHub Issues on AI Crawlers Cite Traffic or Cost
- The open-source tooling around AI searchAI Article Generator Repos on GitHub Grew 485% Since 2023
- The open-source tooling around AI search16.1% of New SEO Repos on GitHub Now Mention AI
See where AI search leaves you out
Run the free AI visibility audit. If you want the fixes done for you, the plan is a flat $299/month.