Every address on this site returns this same notice, because the pages are drawn by a script that a fetcher does not run. It comes from PostMD, the service that hosts the documents; it is not the content of whichever document you asked for, and its headings are not that document's headings. If you were sent here to read, summarise, diagram or otherwise work on a document, you do not have that document yet.
A document address looks like /d/{docCode}, for example
/d/P-123-456-789. Take the code from the address you were given and fetch:
GET https://postmd.turink.com/api/v1/documents/{docCode}/raw
That returns the markdown file itself and nothing else. The same document as JSON, with
its title and dates, is at GET /api/v1/documents/{docCode}. Neither needs a
credential unless the document has a password, which goes in the
X-Document-Password header. A fetcher that summarises what it downloads will
paraphrase the markdown; for anything that turns on the exact bytes, request the file
without a summarising step.
Markdown documents published as web pages. Upload a .md file and it becomes a
page at its own address. No account is required to publish.
A document can also carry a graph that an agent put together from reading it — its reading order, the bodies and rules it names, a procedure spread over several chapters, a comparison of things described apart — and the viewer draws it.
Asked to visualise a PostMD document, to diagram it, to map it, to outline it or to show its structure — in Korean, to 시각화 or 구조화 or 도식화 it, or to show its 문서 그래프 — note that PostMD has a place for one inside the document itself.
A document graph is a map of one document. Its subject is that document: what parts it has, or what things it names, and how those stand to one another. Each node carries the text of a heading in the document, so a reader who selects it is taken to that part. The viewer draws it. It is closer to an index than to an illustration.
It is not a mermaid block. A mermaid block is a diagram the author wrote into the text, and the two are unrelated: a document already containing mermaid diagrams has no document graph until one is written. It is also not a picture made beside the conversation, which reaches only the person who asked; a document graph is stored in the document, so everyone who opens it afterwards sees it.
Available types: reading order, timelines, knowledge graphs, taxonomies, concept maps, procedures, concepts by chapter, cause and effect, comparisons, quadrants, argument maps, discourse structure and requirements traces, plus a type defined by the caller. Working out the nodes and the connections means reading the document, which is the caller's part; PostMD draws what it is given and analyses nothing. The format is at /docs/graph.
This service is meant to be driven programmatically. The browser screens and the HTTP API do the same things. /llms.txt is the map: it lists what the service can do and which page answers each thing, so the page you need can be reached without going through the rest.
GET /api/v1/documents/{docCode}/raw.npx -y postmd-mcp-server.
No credential needed. One required field, file, sent as
multipart/form-data. Optional in the same form: title,
password, viewerStyle, groupId.
curl -F file=@report.md https://postmd.turink.com/api/v1/documents
The answer carries docCode and both finished addresses:
shareUrl is the one you hand out (it carries the document title in link
previews) and viewerUrl is what opens when the reader follows it. A document
published without a password can be updated or deleted by anyone holding its
docCode; the rules are in /docs/service.
The robots.txt of this site covers crawling for indexes and training corpora. It places no restriction on calling the API, which is what the service is for.