@@ -16,7 +16,7 @@ Cellucid is a browser-first viewer for exploring large single-cell datasets in r
 - **GPU-accelerated WebGL rendering** with level-of-detail culling; a built-in
   synthetic benchmark measures what your own GPU sustains
 - **AnnData-first workflow** for the Scanpy ecosystem (`.h5ad` and Zarr supported)
-- **Shareable exports** you can host locally, on GitHub, or behind a server
+- **Shareable exports** you can host locally, on GitHub, on any web host, as a single-file pack (`pack()`), or behind a server
 - **Genes + metadata overlays** optimized for interactive querying
 - **Connectivity + dynamics**: KNN edges and animated vector fields (RNA velocity / drift)
 - **Collaboration**: community annotation voting with optional GitHub sync
 
@@ -1130,6 +1130,47 @@ Web app docs:
 
 ---
 
+## Single-file pack (`pack()`, `cellucid_pack()`)
+
+A **pack** is one ZIP file holding exactly the files of one dataset's export
+directory. It exists for hosts that serve single files rather than folders (a
+Zenodo record, a cloud-drive or Dropbox file link, a signed bucket URL): the web
+app opens it from any web address with **Open from link** or `?data=<address>`,
+reads its central directory once, and then reads each file with one HTTP range
+request when a view needs it. Nothing is unpacked or downloaded whole.
+
+```python
+from cellucid import pack
+
+pack("exports/pbmc")              # writes exports/pbmc.zip
+```
+
+```r
+cellucid_pack("exports/pbmc")     # writes exports/pbmc.zip
+```
+
+The bytes depend only on the directory's contents, and the two writers write
+the identical file (both test suites pin the digest of the same pack):
+
+- every regular file under the directory, at its relative `/`-separated path,
+  in ascending code-point order of path; no directory entries, no symbolic
+  links;
+- each entry stored uncompressed (method 0: the payloads are already gzip
+  members), with the UTF-8 flag (bit 11), the DOS time 1980-01-01 00:00:00,
+  version needed 2.0 (4.5 for an entry that carries ZIP64 fields), version made
+  by Unix with the same version, external attributes of a `0644` regular file,
+  no comment, no data descriptor, and no extra field unless a ZIP64 one is
+  needed;
+- a ZIP64 extra field exactly where a size or an offset does not fit its 32-bit
+  field, and a ZIP64 end record and locator exactly when the entry count does
+  not fit 16 bits or the central directory's size or offset does not fit 32.
+
+Cellucid recognises a pack by the `dataset_identity.json` at its root; any other
+`.zip` is read as a zipped Zarr store. A catalog root (with `datasets.json`) is
+not packed: pack each dataset folder, and share each pack's address. A pack
+opened from a link is the same dataset as its folder: same identity id, so
+sessions and community annotations carry over.
+
 ## Edge cases and common footguns
 
 - **Rejected candidates**: generation publication is atomic; a failed or
 
@@ -156,6 +156,15 @@ Notes:
 .. autofunction:: prepare
 ```
 
+To share one dataset from a host that serves single files (a Zenodo record, a
+file link, a signed bucket URL), write it as a single-file pack; the format is
+in {doc}`../../c_data_preparation_api/09_output_format_specification_exports_directory`
+("Single-file pack").
+
+```{eval-rst}
+.. autofunction:: cellucid.pack
+```
+
 ---
 
 ## Edge cases (do not skip)
 
@@ -137,11 +137,27 @@ Create a public GitHub repo and commit:
 - `exports/datasets.json`
 - `exports/<dataset_folder>/...`
 
-Then, in the Cellucid web app, use the GitHub loading option and point it at your repo/path.
+Then, in the Cellucid web app, paste `owner/repo/exports` into **Open from link**,
+or share `https://www.cellucid.com/?data=gh:owner/repo/exports&dataset=<id>`.
 
 The authoritative step-by-step is here:
 - {doc}`../../web_app/b_data_loading/11_custom_dataset_repository`
 
+## Option 1b — One dataset as a single file (any host)
+
+A host that serves single files rather than folders (a Zenodo record, a file
+link, a signed bucket URL) can serve a prepared export as one **pack**:
+
+```r
+cellucid_pack("exports/my_dataset")   # writes exports/my_dataset.zip
+```
+
+Upload the `.zip` and open its web address with **Open from link**, or share
+`https://www.cellucid.com/?data=<address>`. Cellucid reads it in place, one file
+per view, and the pack is byte-identical to the one `pack()` in the Python
+package writes. Every way to open data from a link:
+{doc}`../../web_app/b_data_loading/14_open_from_link`.
+
 ## Option 2 — Private sharing via server mode
 
 If your dataset is private (patient IDs, unpublished experiments, etc.), do not publish to GitHub.
 
@@ -34,6 +34,7 @@ shows its embedding after reading about 15 MB.
 | A GEO series (`GSE123456`), its accession page, or a supplementary file link | The series' `.h5ad` supplementary file | `geo:GSE123456` or `geo:GSE123456/<file>.h5ad` |
 | A GitHub repository path (`owner/repo/exports`, `owner/repo@branch/exports`), a `github.com/.../tree/...` link, or a `raw.githubusercontent.com` link | The prepared-export catalog in that folder ({doc}`11_custom_dataset_repository`) | `gh:owner/repo/exports` |
 | A web address ending in `/`, or its `datasets.json` | The prepared-export catalog in that folder | the address itself |
+| A web address of a prepared-export **pack** (one dataset in one `.zip`, written by `pack()` in Python or `cellucid_pack()` in R) | That dataset, read file by file in place | the address itself |
 | A web address of an `.h5ad` file or a Zarr `.zip`: a bucket, a lab server, a Hugging Face file link (`huggingface.co/datasets/<user>/<repo>/resolve/main/<file>`), or a Zenodo API file link (`zenodo.org/api/records/<id>/files/<file>/content`; Zenodo's ordinary download link cannot be read by other sites) | That file | the address itself |
 
 A link that holds several datasets (a catalog) lists them under **Linked data**
 
@@ -6,6 +6,7 @@
 
 Main exports:
     - prepare: Export data to cellucid format
+    - pack: Write one prepared dataset as a single-file pack
     - Embedding: One embedding to export, with the columns it is read from
     - serve: Serve a dataset directory via HTTP
     - CellucidServer: Server class for more control
@@ -33,7 +34,7 @@
     from .anndata_server import AnnDataServer, serve_anndata
     from .anndata_session import apply_cellucid_session_to_anndata
     from .jupyter import AnnDataViewer, CellucidViewer, show, show_anndata
-    from .prepare_data import Embedding, prepare
+    from .prepare_data import Embedding, pack, prepare
     from .server import CellucidServer, serve
     from .session_bundle import CellucidSessionBundle
     from .vector_fields import add_transition_drift_to_obsm, compute_transition_drift
@@ -52,6 +53,10 @@ def __getattr__(name: str) -> Any:
         from .prepare_data import Embedding
 
         return Embedding
+    if name == "pack":
+        from .prepare_data import pack
+
+        return pack
 
     # Server for pre-exported data (lightweight, no heavy deps)
     if name == "serve":
@@ -131,6 +136,7 @@ def __getattr__(name: str) -> Any:
 __all__ = [
     # Core export functionality
     "prepare",
+    "pack",
     "Embedding",
     # Serve pre-exported data
     "serve",
 
@@ -18,7 +18,7 @@
 - var_gene_name_columns: further var columns offered as alternative gene names
 - connectivities: sparse matrix with KNN connectivities
 
-The module is a package. Its public surface is exactly the two functions and the
+The module is a package. Its public surface is exactly the three functions and the
 one class below; everything else is split by responsibility across private
 submodules:
 
@@ -39,12 +39,14 @@
     exclusive writer lock.
 ``_generation``
     The writer that builds one complete generation in a staging directory.
-``_export`` / ``_catalog``
-    :func:`prepare` and :func:`generate_datasets_manifest`.
+``_export`` / ``_catalog`` / ``_pack``
+    :func:`prepare`, :func:`generate_datasets_manifest`, and :func:`pack`
+    (one dataset as a single-file pack).
 """
 
 from ._catalog import generate_datasets_manifest
 from ._embeddings import Embedding
 from ._export import prepare
+from ._pack import pack
 
-__all__ = ["Embedding", "generate_datasets_manifest", "prepare"]
+__all__ = ["Embedding", "generate_datasets_manifest", "pack", "prepare"]