library/anydoc

• wasi:filesystem Filesystem read-only

  ** (ro)

act:tools interface — expand a tool for its input schema.

• convert Convert a document to GitHub-Flavored Markdown, preserving headings, lists, tables and footnotes. Handles Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV and PDF. The format is detected from the content; pass `format` for CSV, which carries no signature.

  Convert a document to GitHub-Flavored Markdown, preserving headings, lists, tables and footnotes. Handles Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV and PDF. The format is detected from the content; pass `format` for CSV, which carries no signature.

  Parameter	Type	Description
  data	object | null	Inline document bytes, as a CBOR byte string — or the canonical `{"$bytes": "<base64>"}` envelope over JSON transports.
  path	string | null	Path to a document on the host. Requires a `wasi:filesystem` read grant covering this path.
  format	variant	Parser to use. Omit to detect from the content. Required for CSV, which carries no signature. An explicit value overrides detection.
  filename	string | null	Filename to fall back on for format detection when the content carries no signature. Only the extension is used; content always wins over this, and it is ignored when `format` is given.

  Returns content.

• detect Identify a document's format without converting it. Reports whether the format came from the content signature or only from the filename — an `extension` result on a format that should carry a signature means the file is not what its name claims. Much cheaper than convert.

  Identify a document's format without converting it. Reports whether the format came from the content signature or only from the filename — an `extension` result on a format that should carry a signature means the file is not what its name claims. Much cheaper than convert.

  Parameter	Type	Description
  data	object | null	Inline document bytes, as a CBOR byte string — or the canonical `{"$bytes": "<base64>"}` envelope over JSON transports.
  path	string | null	Path to a document on the host. Requires a `wasi:filesystem` read grant covering this path.
  filename	string | null	Filename to fall back on when the content carries no signature. Only the extension is used. Content always wins over this.

  Returns content.

• extract_assets Extract the images and embedded objects from a document, returned as image content parts with their media type and originating package part. Emits a manifest first, so you can call again with `ids` to fetch only what you need instead of pulling every asset. Not supported for PDF.

  Extract the images and embedded objects from a document, returned as image content parts with their media type and originating package part. Emits a manifest first, so you can call again with `ids` to fetch only what you need instead of pulling every asset. Not supported for PDF.

  Parameter	Type	Description
  ids	array | null	Asset ids to return bytes for. Omit to return all of them. The manifest always lists every asset regardless.
  data	object | null	Inline document bytes, as a CBOR byte string — or the canonical `{"$bytes": "<base64>"}` envelope over JSON transports.
  path	string | null	Path to a document on the host. Requires a `wasi:filesystem` read grant covering this path.
  format	variant	Parser to use. Omit to detect from the content.
  filename	string | null	Filename to fall back on for format detection when the content carries no signature. Only the extension is used; content always wins over this, and it is ignored when `format` is given.

  Returns content.