Publish a static page or site and get its public URL.

Provide exactly one content source. Local paths work only when the server runs locally over stdio; remote (HTTP) callers use html, markdown or files.

List your drops, newest first, with public URLs plus protection and gallery flags.

Your own rows carry their label paths in labels, and any row its description when set.

List your labels in tree order, each with its path, colour and drop count (sublabels counted in).

Check it before labelling a drop and reuse the path a project already has. Labels come into being by labelling a drop; rename, move, recolour and delete them in the dashboard or with the CLI’s dropper labels.

Look up a single drop by slug — name, description, labels, kind, size, protection, public URL.

List the files inside a drop with sizes and content types.

Works on protected drops — owning them grants read-back without a viewer unlock.

Read one file from a drop — fetch current content before editing it with write_drop_file (one file) or update_drop (the whole drop).

Text files return utf8 content; binaries return base64. This returns the stored source: the styled page a browser sees for a .md file exists only at the drop’s /d/ URL. Files over 1 MiB are refused (fetch them via the API instead). Works on protected drops.

Write one file into an existing drop from inline content — create it or replace it in place — leaving every other file untouched.

For small text and assets (HTML, CSS, JS, SVG, JSON, tiny icons): inline bytes travel through the model as tokens, so real images and photos belong to add_drop_asset. To edit text, read_drop_file first and write the whole file back. Nested paths create their directories; the drop’s name, description, labels, password and gallery visibility are unchanged, and a drop left holding more than one file becomes kind "zip". Base64 is checked (standard alphabet, padding) before a request is made; the API caps a file at 20 MB and the drop at 200 MB and 2000 files. A .md file written into a drop renders as a styled page at its own /d/ URL when a browser opens it — table of contents, highlighted code, Mermaid diagrams — while the same URL still returns the stored bytes to anything that does not ask for HTML, and to ?raw=1. To replace a whole drop at once, use update_drop with html or files instead.

Add an image or other binary to an existing drop by having the server download it from a public https URL — the token-free way to put real images in a drop.

The bytes go straight from the source into the drop and never pass through the model (no base64, no tokens). The URL must be a public https:// address that answers 200 with the file itself: redirects are not followed, so give the final asset URL rather than a page, a shortener or a signed-redirect link; 20 MB at most; private, loopback and internal hosts are refused. Creates or replaces exactly one file at path — the served content type comes from its extension — and the drop becomes kind "zip" once it holds more than one file. Reference it from the HTML by relative path, e.g. <img src="images/photo.jpg">; hotlinked images rot when their source changes, ingested copies do not. To replace the whole drop at once, use update_drop with files instead; this tool adds or replaces one file.

Delete one file from an existing drop, pruning any directories it leaves empty; the rest of the drop is untouched.

The drop’s entry page — index.html, index.md or README.md, whichever the drop root serves — cannot be deleted; replace it with write_drop_file instead (a nested index.html is fine). To remove the whole drop, use delete_drop.

Edit an existing drop: rename, change its description or labels, change protection, toggle gallery visibility, or replace content — combinable in one call.

Renaming changes the public URL; changing or clearing the password revokes every viewer unlock. Replacement content uses the same html / markdown / files / path rules as create_drop and replaces the whole drop — to touch one file, use write_drop_file, add_drop_asset or delete_drop_file.

Delete a drop; its public URL stops working immediately and the slug frees up right away.

Recoverable: the drop goes to your trash and stays restorable for 30 days from the dashboard (Profile → Deleted drops) or POST /api/trash/<id>. Only the API’s permanent=1 deletes outright, and this tool never sends it.