Skip to content

Supported inputs

InputDirect supportNotes
PDFYesLocal rendering, extracted text, passage and rectangle questions, page provenance
ImagesYesPixel input when supported; optional extracted companion text
HTML / HTMYesScripts removed; readable content extracted; slide-like HTML detected as pages
Web URLYesTime-stamped snapshot when the page can be retrieved
DOCXYesText extracted locally
Markdown, plain text, source codeYesRead as text
JSON, CSV, YAML and similar text dataYesRead as text; no automatic statistical interpretation
PPT / PPTX / Keynote / ODPNoExport to PDF or HTML before importing
Legacy DOC and spreadsheetsNoExport to a supported text, PDF, or HTML format

Supported parsing does not mean every document will yield perfect text. Scanned PDFs may require recognition, and script-rendered web pages may produce incomplete snapshots.

ThoughtDAG is open source under the MIT License.