The Guild for agents

The Custom Furniture Guild is built to be read by machines as well as people. Every maker, hub, and index is available as structured data an agent can consume directly. Here is the contract.

What we expose

Sitemap (/sitemap.xml) — every indexable URL: makers, category and hub pages, tools.
Feeds (/feeds, and /feeds/[category]) — per-category JSON, one record per maker: location, product types, materials, capabilities, price band, lead time, and the maker's trust tier.
llms.txt (/llms.txt) — a map of the directory for language models: makers by state, hubs by material and product type, and the trust ladder.
llms-full.txt (/llms-full.txt) — the exhaustive per-maker digest.
JSON-LD — every maker, category, and hub page carries schema.org structured data inline.

The trust contract

Every maker carries a trust tier — Listed, Verified, Claimed, Partner, Flagship — exposed on the feeds as trust_tier and partnership_stage. These values are derived from verifiable facts and clamped: a surface can never report a higher tier than the evidence supports. Agents and humans read the same values. Every fact is sourced and dated; numbers are derived, never invented.

How to consume

Crawl /sitemap.xml for the full URL set, fetch /feeds/[category] for structured maker records, and read /llms.txt for the map. Major AI crawlers are explicitly allowed in robots.txt.

What's next

An MCP server over the catalog — so an agent can query the Guild as a tool, not just crawl it — is on the roadmap.