Channel · rank 8 of 100
AI crawlers and assistants
Letting AI assistants read the site and answer about it from the site's own words.
What it is
More and more people ask an AI assistant instead of opening a website. The assistant can only answer well if it may read the site.
TrooIDs lets the AI crawlers in, gives them exactly what a person gets, and hands them the whole knowledge in one file a model can read in one pass: llms-full.txt.
Why it matters
- Rank 8 of 100 in the TrooIDs channel ranking, with a score of 4.5.
- Both models put it in the top class: 80% or more of companies would open it.
- When an AI agent runs this channel, both models agreed it also does the work of: llms.txt files, RSS and Atom feeds, agent instruction files.
- See how the ranking works.
What TrooIDs built
On our test copy of the site, an access report: it states for each of nine AI agents whether the site's rules let it read each page, and lets you test a rule change before it ships.
Proof status, in three parts
- Live on this site: you can check it on trooids.com now.
- Fixture: built and tested with sample data on our test copy of the site; it is not running on trooids.com.
- Architecture: designed or required, not built here yet. Items marked "Owed" are still to do.
Live on this site
- The site's robots.txt lets the AI crawlers read every page that matters, and nothing blocks them by name: 9 of 9 checks each, 21 October 2026.
- Three AI model families answered 7 questions each from the served knowledge files: 21 of 21 correct, 22 October 2026, scored strictly (4 planted wrong answers, 0 accepted).
Fixture
- The access report above: 9 of 9 build obligations met (status 28 October 2026).
Architecture
- Owed: robots.txt also lets crawlers into /api/, so a crawler can start a model call. A rule that keeps crawlers out of /api/ is proposed, not applied.
- Not proven yet: whether assistants actually cite this site.