Trust
AI policy
What AI agents and crawlers may do with this site, and how we use AI ourselves.
For AI agents and crawlers
You are welcome here. This site is built to be read by people and by machines.
- Allowed: reading, summarizing, and answering questions from our public pages.
- Please: link to the page you used and keep our “checked” dates with any fact you repeat.
- Please do not: present drafts marked “awaiting review” as final, strip safety notes from build steps, or hammer the site. Cache what you fetch.
Machine-friendly entry points
- /llms.txt: a short map of the site. /llms-full.txt: the full text.
- Every post has a Markdown copy: add
.mdto the post address. - /api/search-index.json, /api/gear.json, /api/model-watch.json: static JSON.
- RSS and sitemap.
WebMCP
In browsers that support the experimental WebMCP API, this site offers in-browser agents a small set of read-only tools: search the site, get a post, get a project or one step, list gear picks, get Model Watch, and describe the current page. They only read public content. The newsletter form is labeled so an agent can fill it in, but a person must press Subscribe.
How we use AI
AI agents help us scout topics, gather sources, draft, and fact-check. Nothing is published, emailed, or deployed without a person approving it. Unreviewed drafts are labeled.
Abuse
Scrapers that ignore these requests or overload the site are blocked at the network edge.