# LLM Wiki Prompt

A 2025 viral prompt technique based on Andrej Karpathy's ideas, turning Claude into a Zettelkasten-style second brain in Obsidian for personal knowledge management.

The **LLM Wiki Prompt** is a prompt engineering technique that gained viral popularity in 2025, designed to transform a large language model such as Claude into a personal knowledge management system. The technique draws inspiration from the Zettelkasten method, a note-taking system that emphasizes interconnected, atomic notes, and from ideas popularized by computer scientist Andrej Karpathy about using language models as dynamic knowledge tools. Users typically employ the prompt within Obsidian, a markdown-based note-taking application, to generate structured wiki-like notes that link concepts, creating a personal "second brain."

The prompt instructs the model to act as a wiki editor, producing entries with consistent formatting, cross-references, and metadata. It leverages the model's ability to synthesize information and generate internal links, enabling users to build a rich network of knowledge over time. The technique became a notable example of how [large language models](https://www.wikiprompt.org/wiki/large-language-model) can be adapted for productivity and learning, moving beyond simple question-answering to collaborative knowledge creation.

## Origins and Inspiration

The LLM Wiki Prompt emerged in early 2025, gaining traction on social media platforms and developer communities. Its creator, a developer known online as "llion-jones," explicitly credited [Andrej Karpathy](https://www.wikiprompt.org/wiki/andrej-karpathy)'s writings on the future of AI-assisted workflows. Karpathy, a former [OpenAI](https://www.wikiprompt.org/wiki/openai) researcher and Tesla AI director, had discussed the concept of language models as "knowledge engines" that could help users organize and retrieve information more effectively. The prompt's design also reflects principles from the Zettelkasten method, developed by sociologist Niklas Luhmann, which emphasizes creating small, interconnected notes that facilitate creative thinking and discovery.

The original prompt was shared as a template that users could copy into Claude, a chatbot developed by [Anthropic](https://www.wikiprompt.org/wiki/anthropic). It instructed the model to generate wiki-style entries for given topics, complete with a summary, sections, and internal links. The prompt's effectiveness lay in its detailed instructions, which guided the model to produce consistent, well-structured output that could be directly pasted into Obsidian.

## Technical Implementation

The LLM Wiki Prompt typically includes a system message that defines the model's role and output format. For example, the prompt might say: "You are a wiki editor. Write an encyclopedia article on the given topic. Use neutral tone, include specific facts, and link to related concepts using double brackets." The prompt also specifies the desired structure, such as a short summary, several sections, and a list of categories.

Users often customize the prompt to suit their needs, adding instructions for tone, length, or specific fields. The output is designed to be compatible with Obsidian's markdown syntax, which supports wiki links and tags. This allows users to create a network of notes that can be navigated and searched, mimicking the functionality of a personal wiki.

One key aspect is the use of [attention mechanisms](https://www.wikiprompt.org/wiki/multi-head-attention) in the underlying model, which enables it to identify relationships between concepts and generate relevant links. The prompt leverages the model's training on vast corpora of text, including encyclopedic sources, to produce accurate and informative entries.

## Applications and Use Cases

Users have applied the LLM Wiki Prompt in various domains, including academic research, software development, and personal productivity. For instance, a researcher might use it to create notes on papers, linking them by topic and methodology. A developer could generate documentation for code libraries, with links to related functions and modules. The technique is particularly valued for its ability to surface connections between disparate ideas, fostering serendipitous discoveries.

In education, teachers have used the prompt to create study guides that link concepts across subjects. In business, professionals have used it to build knowledge bases for projects, capturing insights and references. The prompt's flexibility allows it to be adapted for any field where structured knowledge is valuable.

## Reception and Impact

The LLM Wiki Prompt received widespread attention in 2025, with many users sharing their customized versions and workflows. It was praised for making advanced knowledge management accessible to non-experts, requiring no programming skills. Critics noted that the quality of output depends on the underlying model's knowledge and the user's ability to refine prompts. Some also raised concerns about over-reliance on AI-generated content, emphasizing the need for human verification.

Despite these caveats, the technique has influenced subsequent prompt engineering practices, highlighting the potential of [generative AI](https://www.wikiprompt.org/wiki/generative-ai) for personal knowledge management. It also contributed to the growing ecosystem of tools that integrate language models with note-taking applications, such as Obsidian plugins and [Anthropic](https://www.wikiprompt.org/wiki/anthropic)'s API.

## Future Directions

As language models continue to improve, the LLM Wiki Prompt is likely to evolve. Future iterations might incorporate multimodal capabilities, allowing users to generate notes from images or audio. Integration with [AI](https://www.wikiprompt.org/wiki/artificial-intelligence) agents could enable automated knowledge curation, where the model proactively suggests new notes or updates existing ones. The technique may also be adapted for collaborative use, enabling teams to build shared wikis with AI assistance.

The prompt's success underscores a broader trend: the use of [large language models](https://www.wikiprompt.org/wiki/large-language-model) not just as answer engines but as partners in knowledge work. As tools like Obsidian and [Claude](https://www.wikiprompt.org/wiki/claude) become more intertwined, the line between human and machine-generated knowledge may blur, raising questions about authorship and originality. Nonetheless, the LLM Wiki Prompt represents a practical, creative application of AI that empowers individuals to manage information more effectively.

---
Source: https://www.wikiprompt.org/wiki/llm-wiki-prompt
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-10-08T07:37:30.939633+00:00