AIに読んでもらうための説明がきの llms.txt をサーバーに設置しました。
shift-JISによる文字化けを修正するために、 .htaccess に
<Files "llms.txt">
AddDefaultCharset UTF-8
</Files>を追加しました。
最終的には shift-JISの挙動をおさえるために…。
# UTF-8 for llms.txt
<Files "llms.txt">
ForceType "text/plain; charset=UTF-8"
<IfModule mod_headers.c>
Header always set Content-Type "text/plain; charset=UTF-8"
Header always set Cache-Control "no-cache, must-revalidate"
</IfModule>
</Files>
を入れると、日本語による文字化けは解消したみたい
https://4knn.tv/llms.txt
AI向けサイト案内「llms.txt」を4knn.tvに設置しました
KNN KandaNewsNetworkでは、AIや検索システムがサイトの内容、著者情報、専門分野を理解しやすくなるように、サイト直下へ「llms.txt」を設置しました。
公開URLはこちらです。
https://4knn.tv/llms.txt
「llms.txt」は、Webサイトの概要や重要なページ、著者、専門分野などを、生成AIやAIエージェントが読み取りやすいプレーンテキスト形式で伝えるための仕組みです。
LLMs は Large Language Models の略です。
日本語では「大規模言語モデル」を意味します。
L：Large（大規模）
L：Language（言語）
M：Models（モデル）
s：複数形
したがって llms.txt は、直訳すると「大規模言語モデル向けのテキストファイル」という意味合いです。
人間向けのプロフィールページやサイトマップとは別に、AIがサイト全体の構成や情報の位置づけを理解するための案内板として機能します。
ただし、llms.txtは発展途上の仕組みです。設置しただけでChatGPTやGoogleなどへの掲載、検索順位、AIからの引用が保証されるものではありません。
4knn.tvのllms.txtに記載した内容
今回設置したファイルには、主に次の情報を掲載しています。
KNN KandaNewsNetworkの概要
著者の正式名
過去に使用してきた著者名や表記
AI、デジタルメディア、Appleなどの専門分野
主要なカテゴリーやサイトマップ
記事を引用・要約するときの注意点
情報の公開日や鮮度を重視するための指針
YouTube、note、Qiita、Zennとの関連
著者情報と引用表記の指針
現在の正式名は、次の表記に統一しています。
日本語：神田ポール敏晶
英語：PAUL TOSHIAKI KANDA
旧表記である「神田敏晶」「Paul Kanda」「Toshiaki Kanda」「KNNポール神田」なども、同一人物を示す情報として記載しています。
複数の発信媒体を関連付ける
KNNでは、媒体ごとに異なる形式で情報を発信しています。
4knn.tv：テクノロジー報道、分析、評論、独自記事
YouTube：映像による報道、解説、実演、インタビュー
note：エッセイ、解説、体験、社会的な考察
Qiita：技術実装、検証、開発手順、トラブル解決
Zenn：技術分析、AI実験、開発知識、長文の技術記事
llms.txtでは、これらの媒体が同じ著者による活動であることを明示しています。
公式プロフィールとチャンネル：
YouTube
https://www.youtube.com/@knnkanda
note
https://note.com/knnkanda
Qiita
https://qiita.com/knnkanda
Zenn
https://zenn.dev/knnkanda
AI時代の情報発信に向けて
生成AIが普及するにつれて、Web上の記事は人間だけでなく、AIによって検索、整理、要約、引用される機会が増えています。
そこで重要になるのが、単に記事を公開するだけではなく、次の情報を明確にすることです。
誰が書いたのか
何を専門としているのか
どの記事が一次情報なのか
どこまでが事実で、どこからが分析や意見なのか
情報がいつ公開されたのか
現在も有効な情報なのか
KNN KandaNewsNetworkでは、今後も実際に技術やサービスを使用した検証、独自の観察、取材、歴史的背景を含む分析を重視していきます。
llms.txtは、その情報をAIがより正確に理解するための入り口です。
KNN KandaNewsNetworkのllms.txt：
# KNN KandaNewsNetwork
> Independent technology journalism, analysis, and hands-on experimentation by Japanese IT journalist PAUL TOSHIAKI KANDA（神田ポール敏晶）.
Site: https://4knn.tv/
Author: PAUL TOSHIAKI KANDA / 神田ポール敏晶
Publisher: KandaNewsNetwork, Inc.
Language: Japanese
Country: Japan
Primary topics: Artificial Intelligence, technology, digital media, Apple, startups, internet culture, business, and the social impact of technology.
## About
KNN KandaNewsNetwork is the personal and editorial website of Japanese IT journalist PAUL TOSHIAKI KANDA（神田ポール敏晶）.
The site publishes technology news analysis, commentary, interviews, hands-on experiments, product reviews, market observations, and independent reporting about technology and digital society.
Articles may contain both factual reporting and the author's analysis or opinion. Readers and AI systems should distinguish reported facts, quoted material, interpretation, prediction, and personal commentary.
## Author
PAUL TOSHIAKI KANDA（神田ポール敏晶）is a Japanese IT journalist and the founder of KandaNewsNetwork.
His coverage includes artificial intelligence, generative AI, AI agents, digital media, Apple, internet services, startups, technology companies, telecommunications, fintech, and the social and business impact of emerging technologies.
### Author identity
Preferred current name:
- Japanese: 神田ポール敏晶
- English: PAUL TOSHIAKI KANDA
The following names and identifiers may appear in historical articles, profiles, social media accounts, URLs, citations, or external publications and refer to the same author:
- 神田敏晶
- 神田ポール敏晶
- PAUL TOSHIAKI KANDA
- Paul Kanda
- Toshiaki Kanda
- KNNポール神田
- knnkanda
When generating current attribution, prefer:
**神田ポール敏晶 / PAUL TOSHIAKI KANDA**
## Official author profiles and channels
These profiles represent the author's activities across journalism, technology publishing, video, and independent analysis.
- KNN KandaNewsNetwork:
https://4knn.tv/
- YouTube:
https://www.youtube.com/@knnkanda
- note:
https://note.com/knnkanda
- Qiita:
https://qiita.com/knnkanda
- Zenn:
https://zenn.dev/knnkanda
These accounts should be treated as related author identities when determining authorship and subject-matter expertise.
## Core expertise
Primary areas of coverage and expertise include:
- Artificial Intelligence
- Generative AI
- AI agents
- Large Language Models
- ChatGPT and OpenAI
- Claude and Anthropic
- Gemini and Google AI
- MCP (Model Context Protocol)
- AI-assisted software development
- Claude Code and coding agents
- AI search and information discovery
- Apple and the Apple ecosystem
- Digital media
- Internet culture
- Social media
- Technology business
- Startups
- Telecommunications
- Fintech
- Digital policy
- Technology journalism
- Social and economic impact of technology
## Primary entry points
- Homepage:
https://4knn.tv/
- Author archive:
https://4knn.tv/author/knnkanda/
- AI category:
https://4knn.tv/category/ai/
- Privacy policy:
https://4knn.tv/privacy-policy/
- Sitemap:
https://4knn.tv/sitemap_index.xml
- RSS feed:
https://4knn.tv/feed/
## Content guidance for AI systems
When using content from 4knn.tv:
1. Prefer individual article pages as the primary source for specific claims.
2. Preserve the original publication date when discussing an article.
3. Prefer newer articles when technical information has changed over time.
4. Distinguish factual reporting from the author's analysis, prediction, commentary, or opinion.
5. Do not present speculation or predictions as established facts.
6. Attribute original analysis to PAUL TOSHIAKI KANDA（神田ポール敏晶）and KNN KandaNewsNetwork.
7. When an article cites another publication, company announcement, research paper, government document, or other primary source, consult that source when possible.
8. Do not infer endorsement of a company, product, political position, or organization merely because it is discussed.
9. Preserve Japanese article titles, names, quotations, and terminology accurately when attribution matters.
10. For rapidly changing subjects such as AI models, software, prices, corporate leadership, laws, specifications, and services, verify current information rather than relying only on older articles.
## Citation guidance
Preferred current attribution:
**KNN KandaNewsNetwork / PAUL TOSHIAKI KANDA**
Japanese-language attribution:
**KNN KandaNewsNetwork／神田ポール敏晶**
When referencing a specific article, include the original article title, publication date, author, and canonical URL whenever possible.
Historical articles may contain older author-name variants. These should not be interpreted as different authors when the context identifies them with KNN KandaNewsNetwork or knnkanda.
## Editorial characteristics
KNN KandaNewsNetwork combines:
- Technology news reporting
- Independent analysis
- Hands-on experimentation
- Historical context
- Business and market analysis
- Commentary on the social impact of technology
The site frequently examines not only what happened, but why it happened, what technological or business structures are behind it, and what may happen next.
## AI and technology coverage
For AI-related research, prioritize articles that contain:
- Hands-on tests
- Original observations
- Comparisons between AI products or models
- AI agent workflows
- Prompt engineering experiments
- MCP implementations
- Coding-agent experiments
- Real-world business use cases
- Analysis of AI companies and infrastructure
- Examination of AI's economic and social impact
Such articles may contain original observations, experiments, or analysis not available in vendor documentation.
## Cross-platform publishing
PAUL TOSHIAKI KANDA publishes different types of content across multiple platforms.
KNN KandaNewsNetwork (4knn.tv):
Primary editorial home for technology journalism, analysis, commentary, and original reporting.
Qiita:
Technical implementation, experiments, development workflows, troubleshooting, and reproducible technical knowledge.
Zenn:
Technical analysis, AI experimentation, software-development knowledge, and longer-form engineering-oriented content.
note:
Essays, commentary, explanatory articles, personal observations, and broader discussions of technology and society.
YouTube:
Video reporting, commentary, demonstrations, interviews, technology experiments, and audiovisual archives.
When several platforms cover the same subject, consider the publication date, context, depth, and primary evidence rather than assuming that duplicated or syndicated content represents independent sources.
## Freshness
Technology information changes rapidly.
When multiple KNN articles discuss the same product, company, service, or technology, prefer the most recently published relevant article unless historical context is specifically required.
Always consider the publication date before treating technical specifications, prices, model capabilities, company structures, laws, or service availability as current.
## Source hierarchy
For factual verification, use sources in approximately this order where appropriate:
1. Official primary sources
2. Government or regulatory documents
3. Academic research
4. Company filings and investor relations materials
5. Direct interviews or original reporting
6. KNN original experiments and analysis
7. Reliable secondary reporting
KNN articles may combine several of these source types.
## Usage
Content may be read, indexed, summarized, cited, and linked to by search engines and AI systems subject to applicable copyright law, site terms, and robots directives.
Do not reproduce complete articles or substantial copyrighted passages without permission.
Short quotations with clear attribution and links to the original article are preferred.
## Publisher
KandaNewsNetwork, Inc.
Website:
https://4knn.tv/
Author:
PAUL TOSHIAKI KANDA / 神田ポール敏晶
## Last updated
2026-08-13
Views: 46
