Skip to main content
GET
Preview. This endpoint is available to organisations in the Wordsmith MCP and files pilot. Other organisations receive 403 mcp_not_enabled. See the preview overview.
Returns one page of a file’s plain text. Wordsmith extracts the text from documents, spreadsheets, presentations, images and other supported types. Tracked changes in a Word document are kept. Read a long file page by page:
  1. Call without offset. The first page starts at character 0.
  2. While next_offset is not null, call again with offset set to next_offset.
  3. When next_offset is null, you have the last page.
total_chars is the length of the whole text, so you can tell how many pages are left. An offset at or past the end returns empty text and a null next_offset. A file is served only once its malware scan is clean. While the scan is still running, the response is 409 scan_in_progress; retry shortly. A file the scan flagged returns 404, as a missing file does.

​
Authentication

Send an API key, or an OAuth access token with the files:read scope, as a bearer token in the Authorization header. files:write does not include files:read. A token without the scope gets 403 with "error": "insufficient_scope".

​
Path parameters

​
string (uuid)
required
The file’s id.

​
Query parameters

​
integer
default:"0"
Character offset to start the page at. 0 or greater.
​
integer
default:"20000"
The most characters to return. 1 or greater. Values over 100,000 are treated as 100,000.

​
Response

​
string (uuid)
The file’s id.
​
string
This page of the text.
​
integer
Character offset of the first character in text.
​
integer | null
The offset to read the next page from. null on the last page.
​
integer
The number of characters in the whole text.

​
Errors

See Errors for the error format.