# 🤖 Discourse llms.txt Generator Plugin **URL:** https://meta.discourse.org/t/discourse-llms-txt-generator-plugin/388170 **Category:** Plugin **Tags:** ai, security-issue **Created:** [November 11, 2025, 6:45pm UTC](https://meta.discourse.org/t/discourse-llms-txt-generator-plugin/388170 "2025-11-11T18:45:19Z") **Posts on this page:** 1 **Showing post:** 2
### Author: ![Ivan\_Rapekas](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/ivan_rapekas/32/248924_2.png) [@Ivan\_Rapekas](https://meta.discourse.org/u/Ivan_Rapekas) #### Post date: [November 20, 2025, 3:25pm UTC](https://meta.discourse.org/t/discourse-llms-txt-generator-plugin/388170/2 "2025-11-20T15:25:59Z")
I installed the plugin a couple of days ago. The installation was smooth, thanks to author’s documentation. What I see now. It were generated a lot of txt’s at my site, for example `https://discuss.rabkesov.ru/llms.txt`. I now that it is too early to want to see some AI chatbot (Perplexity) shows my content as an answer in chat. So, to check an ability to be suitable for LLM, I uploaded it to Qwen (unfortunately it doesn’t support the link above): ```plaintext Could you analyze the page llms.txt and tell me what you understood from what you have read? Make a tree in the form of a mind map (Mermaid) ``` > P.S. If I ask to translate everything into English, the tree breaks. ![image](https://global.discourse-cdn.com/meta/original/4X/1/c/6/1c6c71edeb0ee1189c2502ec8de833246011e288.png) ![image](https://global.discourse-cdn.com/meta/original/4X/3/a/9/3a925165f75b59743183c65fe80f1c7370d91114.png) Here it is a structure of my site (the **public part** only). LLM could created it from the `llms.txt`! I think it is only a small part of possible cases. * * * > **Text output without Mermaid** > > ```plaintext > Could you analyze the page llms.txt and tell me what you understood from what you have read? > > ``` > > Yes. Based on the content of `llms.txt`, here’s what I understood: > > The file describes **“discuss | Комната ломки”** , which translates roughly to _“Discussion Forum | The Breaking Room”_ — a space metaphorically intended for breaking _illusions_ in IT, not people. Its tagline and focus suggest a reflective, critical, and community-driven environment where IT professionals can examine assumptions, share experiences, and collaborate on practical challenges. > > ### Core Themes & Purpose: > > - **Critical reflection on IT practices** , career paths, organizational dynamics, and personal growth. > - Emphasis on **tools and methodologies** that simplify work: documentation (Markdown/GitLab/Docsify), visualization (Mermaid/Drawio), feedback mechanisms, and collaborative drafting — particularly within a **Discourse-based corporate platform**. > - A mix of **technical, philosophical, and personal topics** , often intersecting: e.g., “Why we choose easy tasks”, “Cost of peace of mind”, “Procrastination and decision avoidance”. > > ### Structure: > > The site is organized into thematic **categories** , including: > > - **AI** : Discussions on models (e.g., Gemma, LM Studio), deployment (Runpod, GPU calculators), and tooling. > - **Discourse** : Technical backend topics (site configuration, rake tasks, auth methods, hidden parameters). > - **Основная (Main)**: Core reflections on work culture, leadership, documentation standards, and responsibility. > - **Блог (Blog)**: Personal narratives (e.g., career shifts, relocation to Samara, workplace incidents). > - **Школа (School)**: Resources for students, including ethics (parental consent for 14+ users) and digital literacy (UTM parameters). > - **Анонимная (Anonymous)**: Safe space for sensitive ideas. > - **Книги (Books)**, **Мысль дня (Thought of the Day)**, etc. — supporting intellectual and daily reflection. > > ### Notable Technical & Cultural Signals: > > - **Multilingual & international audience** : e.g., parallel Russian/English authentication guides. > - Concern for **inclusive authentication** : e.g., preserving Google/GitHub login for non-Russian users, while also supporting domestic providers (VK, [Mail.ru](http://Mail.ru), OK). > - Strong interest in **automation, standardization, and DevOps** : GitLab CI/CD, artifact handling, Runpod integration, SSH-based deployment. > - Recurring theme of **professional disillusionment** , especially among senior engineers moved into support roles — and strategies to cope or plan exits. > > In summary, `llms.txt` outlines a mature, thoughtfully designed community platform that blends technical depth with psychological and organizational introspection — targeting experienced IT practitioners navigating complexity, burnout, and the search for meaning in their work. > > Let me know if you’d like a visual map, thematic clustering, or suggestions for next steps (e.g., improving discoverability, engagement, or onboarding). * * * As for me, It is a great plugin to be used in SEO etc. Some point to explore: 1. How to handle non-English content? Is it matter to LLM the EN-language of text in `llms.txt`? 2. Find the proof that the final root llms.txt never contains private pages (`llms.txt` is created for **every** page of the site at this moment) 3. Check permissions at `/admin/site_settings/category/security?filter=crawler` so that necessary AI bots were not blocked 🙂 4. Thoughts about occasional mixing site structure with a memories, which is saved in LLM previously (this is _not a plugin fault_, but a chatbot feature, see example below) About the point #4. The first time Qwen has decided that: - I planned to leave my work - I had troubles in relationship 🙂 and **added** these points into site description! Qwen **may** mix memories with a current request of analyzing the sitemap, provided as `llms.txt`. Be aware 🙂 --- _[View the full topic](https://meta.discourse.org/t/discourse-llms-txt-generator-plugin/388170)._