Developers

Every record is also published as structured data, with its rules’ basis and claim keys intact, so a model reads the distinctions instead of flattening them.

Machine-readable records

Works today

One JSON file per Aesthetic in the collection, at data/v1/<slug>.json: every claim with its basis and the sources that support or dispute it, plus the values each page is built with, kept separate. 73 files, listed in data/v1/index.json.

data/v1/swiss-style.jsondata/v1/index.json

Agent context

Works today

A Markdown style guide per Aesthetic in the collection, written for coding agents. 73 files.

prompt/swiss-style.md

Exports

Works today

The whole catalogue for search, and every mixable record with its rules.

search-index.jsonmix-data.json

JSON schema

Not built

The V2.1 schema bundle lives in the repository. It is not published on the site.

API

Not built

Query records, Elements (Modules, in the schema) and their links over HTTP. Not built; the static files above are the whole machine surface today.

MCP server

Not built

Let an agent ask Aesthetary directly instead of reading a pasted file. Not built.

Structured Recipes

Not built

Recipes as agent context. Waits on Recipes existing.

Evaluation

Not built

Grade generated work against a language programmatically, using the rubric the Build pages already carry.

How the machine layer is built · Licensing

Feedback