Nakshora for AI & LLMs · Nakshora Nakshora 3.1 Docs

Nakshora is AI-ready: llms.txt, a complete LLM documentation dump, a structured utility corpus and an SFT dataset — teach any model the framework.

Nakshora was the first utility framework built with LLMs as a first-class audience. Nakshora 3.1 ships four machine-readable artifacts:

Artifact	What it is	Use it for
`llms.txt`	Concise, spec-compliant framework summary	Agent context, search grounding
`llms-full.md`	Complete documentation in one Markdown file	Long-context RAG, offline assistants
`ai/corpus.json`	Structured corpus of every utility (11,417 entries)	Embeddings, tool-use catalogs
`ai/sft-train.jsonl`	Instruction-tuning pairs	Fine-tuning code models

Give an agent the framework in one line#

text

Add to your agent instructions:
"Read https://raw.githubusercontent.com/nakshora/nakshora/main/llms.txt before generating Nakshora markup."

Export the corpus yourself#

terminal

bash

npx nakshora export:ai --out corpus.json

The corpus shape: { class, css, description, category, example, responsiveExamples[] } for every utility — perfect for embeddings or function-calling catalogs.

Note. This very documentation platform publishes its own llms.txt and per-version llms-full.md — see the site footer.