The public API exposes task descriptions as description / markdown_description on read and description / markdown_content on write. Markdown is lossy in both directions:

Reading flattens toggles, banners, columns, table formatting, and inline mentions. Synced Content disappears entirely.

Writing can't produce any of those blocks. The markdown importer also inserts extra paragraphs in nested lists.

Synced Content blocks have no endpoints at all.

What I'm asking for

A documented read/write field for structured description content on the task endpoints. The Quill { ops } format the web client already uses would be fine. Plus read/update endpoints for Synced Content blocks.

A minimal version would go a long way: the v2 task endpoints already accept description: { ops: [...] } today, because that's what the web app sends. Documenting that behavior and committing to its stability would unblock most of this without new endpoints.

I maintain an open-source ClickUp CLI (krodak/clickup-cli). I just had to turn down contribution related to this: https://github.com/krodak/clickup-cli/pull/144

I declined it as it relies on web app's private endpoints and a session JWT scraped from DevTools.