Documentation for LLMs
The Snowplow documentation is available in formats designed for LLMs, so your AI assistant can find and read the pages it needs.
If your assistant is connected to the Snowplow MCP server, it can also fetch documentation pages through the server's documentation tool, alongside working with your Snowplow account.
Documentation index in llms.txt
This documentation follows the llms.txt standard, providing structured information to help an LLM use the site.
An index is available at llms.txt.
An extended version is also available at llms-full.txt, that includes the complete text of the current pages. For token efficiency, sections relating to older versions of components aren't included in this file. These sections are still listed in the llms.txt index, labeled as [previous version].
The llms-full.txt file is very large. It might be more effective to access individual pages as needed, using the Markdown access method described below.
Documentation pages as Markdown
Every documentation page is available as Markdown. To download a page's content, use the Download or Copy Markdown buttons above the page title.
Following the llms.txt standard, you can access the Markdown page directly by changing the trailing / in the URL to .md. For example:
- HTML:
https://docs.snowplow.io/docs/signals/concepts/ - Markdown:
https://docs.snowplow.io/docs/signals/concepts.md
Let your LLM know that this format is available, so it can retrieve content efficiently.
Tutorials as single pages
Each tutorial is also available as a single Markdown file containing all of its steps in order. Tutorials have no page of their own on the site, so these files don't follow the URL pattern above. Append .md to the tutorial's directory path instead:
- Tutorial step:
https://docs.snowplow.io/tutorials/signals-quickstart/start/ - Complete tutorial:
https://docs.snowplow.io/tutorials/signals-quickstart.md
The combined file opens with a list of the tutorial's steps and their URLs, followed by the content of each step. The llms.txt index links to these files rather than to individual steps, so an LLM retrieves a whole tutorial in one request.