Problem: docs.nextcommerce.com has no llms.txt (404), while www.nextcommerce.com and
developers.nextcommerce.com both serve one. The merchant docs portal is the surface an
evaluating agent most needs an index for, and it's the only public portal missing one.
Example: curl -s -o /dev/null -w '%{http_code}' https://docs.nextcommerce.com/llms.txt
→ 404 (checked 2026-08-17). developers.nextcommerce.com/llms.txt → 200.
What we see: AI agents are actively reading our surfaces — Cloudflare zone analytics for
the week of Aug 11–17 show ~23k requests from AI crawlers (ChatGPT-User 1,447, ClaudeBot 828,
OAI-SearchBot 707, PerplexityBot 417, Claude-User 219). ChatGPT-User and Claude-User are
live, user-driven evaluation sessions, not index crawlers. In a structured evaluation test
(three frontier agents asked to assess NEXT from our public surfaces), agents that found the
developers llms.txt used it heavily; one flagged the docs 404 explicitly.
Suggested shape: mirror the developers portal approach — a generated index of the docs
tree with one-line descriptions. Fumadocs sites can emit this from the content manifest at
build time.
Who's affected: every prospect or merchant whose agent evaluates or answers questions
from merchant docs; support deflection quality for agent-mediated questions.
Problem: docs.nextcommerce.com has no llms.txt (404), while www.nextcommerce.com and
developers.nextcommerce.com both serve one. The merchant docs portal is the surface an
evaluating agent most needs an index for, and it's the only public portal missing one.
Example:
curl -s -o /dev/null -w '%{http_code}' https://docs.nextcommerce.com/llms.txt→ 404 (checked 2026-08-17). developers.nextcommerce.com/llms.txt → 200.
What we see: AI agents are actively reading our surfaces — Cloudflare zone analytics for
the week of Aug 11–17 show ~23k requests from AI crawlers (ChatGPT-User 1,447, ClaudeBot 828,
OAI-SearchBot 707, PerplexityBot 417, Claude-User 219). ChatGPT-User and Claude-User are
live, user-driven evaluation sessions, not index crawlers. In a structured evaluation test
(three frontier agents asked to assess NEXT from our public surfaces), agents that found the
developers llms.txt used it heavily; one flagged the docs 404 explicitly.
Suggested shape: mirror the developers portal approach — a generated index of the docs
tree with one-line descriptions. Fumadocs sites can emit this from the content manifest at
build time.
Who's affected: every prospect or merchant whose agent evaluates or answers questions
from merchant docs; support deflection quality for agent-mediated questions.