docs-for-llms

This repo contains single file concatenated documentation for every released version of LLM and several other tools.

It's designed to work with the llm-docs plugin:

llm install llm-docs
llm -f docs:llm 'how do I embed a binary file?'

Asking more questions about the same documentation - or asking follow-up questions using llm -c or even llm chat -c - will benefit from OpenAI's token cache pricing.

Projects represented here:

• datasette (this is a large one)
• datasette-plugins is the subset of that documentation needed for writing Datasette plugins
• llm
• s3-credentials
• shot-scraper
• sqlite-utils

Note that some of these files are a little large though:

curl -s 'https://raw.githubusercontent.com/simonw/docs-for-llms/refs/heads/main/datasette/1.0a16.txt' \
  | ttok

152131

That's 152,000 tokens - too large for gpt-4o-mini and potentially quite expensive to process.