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-pluginsis 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' \
| ttok152131
That's 152,000 tokens - too large for gpt-4o-mini and potentially quite expensive to process (but fits easily within the 1 million token limits of both the Gemini and the GPT-4.1 models).