Reading this site from a program

ComposerWire has no accounts, keys or write access. What a program can use is the site itself, read with plain GET requests.

Every page has a Markdown copy. Send Accept: text/markdown to any page's address and you get Markdown instead of HTML, with the same text under the same rules.

The sitemap lists every page. Listing pages are under /listings/, issue pages under /issues/. A listing's status is worked out from the date of the nightly build, so a listing closes on the site without any change to its text.

robots.txt says how the content may be used, in its Content-Signal line.

The same is described in OpenAPI at /openapi.json, and catalogued at /.well-known/api-catalog under RFC 9727. The front page points to all three in its Link headers.

/auth.md says, in the auth.md format agents read for registration, that there is nothing to register for and no credential to send.

There is no search endpoint, no feed of listings as data, and no health endpoint.