Autype

Return a Markdown chunk by outline chunk ID or line range. This retrieval is based on Extended Markdown lines, not JSON sections.

documents_get_markdown_chunk

Return a Markdown outline with heading chunks, directive hints, line ranges, hashes, and document revision. Use this before chunk retrieval or patching long documents.

documents_get_markdown_outline

List persistent documents in the organization. Use documents_get_markdown to read one in the preferred AI-friendly format.

documents_list

Patch a persistent document's Extended Markdown by line ranges without making the agent rewrite the whole document.

documents_patch_markdown

Queue an export of the latest persistent document snapshot using existing automation credits. Requires a workspace-visible project; the source stays unchanged. Returns a job ID; poll with exports_get_status. Do not resubmit after an uncertain timeout: a duplicate job consumes credits again.

documents_render

Update a persistent document from Extended Markdown. Omitted metadata is preserved by the API.

documents_update_markdown

Inspect an imported DOCX as validated Autype document JSON. First use files_import_attachment for the attached Word file. Preserves supported text, tables, styles and embedded images; does not render or save an editable document. For direct PDF conversion use docx_to_pdf. This ChatGPT surface does not expose a raw JSON roundtrip renderer.

docx_import

Convert an uploaded Word document to PDF. Starts a credit-consuming job; input remains unchanged.

docx_to_pdf

Validate and queue a temporary document export. Consumes existing automation credits; does not save an editable document. Returns jobId; poll exports_get_status. Do not retry submission after an uncertain timeout without checking jobs.

exports_create

Read a render job's status and signed download link. No rendering or credit charge. Links are temporary.

exports_get_status

Display an export and preview one PDF page without consuming credits. PDFs up to 8 MB can be previewed; other formats or larger files retain their download link. No additional render job is started.

exports_view

Get toolkit file metadata. Binary content is intentionally not returned to the LLM.

files_get

Import an explicitly attached ChatGPT PDF, DOCX, ODT, HTML, Markdown or image into temporary Autype storage (50 MB maximum, expires in 60 minutes). No OCR or paid job is started.

files_import_attachment

List non-expired toolkit files.

files_list

Import an attached PNG, JPEG or WebP logo/image for rendering (25 MB maximum). Returns refPath; temporary images expire after 24 hours unless referenced by a persistent document.

images_import_attachment

Convert an uploaded OpenDocument file to PDF. Starts a credit-consuming job; input remains unchanged.

odt_to_pdf

Compress an uploaded PDF into a new file. Starts a credit-consuming job; input remains unchanged.

pdf_compress

Merge 2–20 uploaded PDFs in the specified order into a new output. Starts a credit-consuming job; inputs remain unchanged.

pdf_merge

Split an uploaded PDF into page ranges such as ['1-3','4-6']. Creates new outputs and consumes credits; input remains unchanged.

pdf_split

List all projects in the organization. Results are paginated and sorted by creation date (oldest first). Each project contains an `id`, `name`, and `visibility`. Use documents_list with the returned project ID to find documents, then documents_get_markdown or documents_get_markdown_outline to read a selected document.

projects_list

Get one workspace or system style, including its full defaults/document definition. System IDs use the explicit `system:<id>` namespace and are immutable; they can be passed to temporary render tools as style_preset_id.

styles_get

List compact reusable style summaries. Every item has source=workspace or source=system and an editable flag, so owned styles and built-in templates are never ambiguous. Always call styles_get before applying a selected style.

styles_list

Fill and export an organization-owned template using existing automation credits, leaving the source unchanged and creating no editable document. Returns a render job ID; use exports_get_status until it completes. Built-in catalog templates are rejected. Do not resubmit after an uncertain timeout: a duplicate job consumes credits again.

templates_fill_and_render

Get one accessible template as Extended Markdown plus metadata. The response includes scope and productive; only productive workspace templates may be passed to templates_fill_and_render.

templates_get_markdown

List compact, paginated template summaries from exactly one scope. Productive work defaults to organization-owned workspace templates; use catalog only for optional inspiration. The two collections are never mixed.

templates_list_markdown

Get toolkit job status, result metadata, output file IDs, and download URL.

tools_get_job