- From: Mike Gifford via GitHub <noreply@w3.org>
- Date: Thu, 12 Mar 2026 19:26:28 +0000
- To: public-webai@w3.org
I am experimenting. I've shared some of these on Slack, but: What if we defined our expectations, much like a AGENTS.md file https://mgifford.github.io/ACCESSIBILITY.md/ I've found this works reasonably well, when for instance I want to implement dark mode, and want to point to an example like: https://github.com/mgifford/ACCESSIBILITY.md/blob/main/examples/LIGHT_DARK_MODE_ACCESSIBILITY_BEST_PRACTICES.md Why don't we give the bots a list of those we trust (and those who are willing to have their sites crawled): https://github.com/mgifford/ACCESSIBILITY.md/blob/main/examples/TRUSTED_SOURCES.yaml I got this idea from Ted Drake: https://github.com/7mary4/a11y-ai-training If you're logged into Google in your browser, try going to the and adding a URL: https://gemini.google.com/gem/1-oRtNsS6CLM1WMI3-JJ9bn_Se99us5sJ I've definitely got more ways that this can be an opportunity for our industry. There are obviously a lot of risks. Lots of certainty too in terms of the dangers of data centers. This is a very sharp double-edged sword. -- GitHub Notification of comment by mgifford Please view or discuss this issue at https://github.com/w3c/webai/issues/14#issuecomment-4049437055 using your GitHub account -- Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config
Received on Thursday, 12 March 2026 19:26:29 UTC