NoteCapsule · for Jupyter Notebook & Google Colab

Never lose your Jupyter or Colab notebook again.

NoteCapsule creates reproducible project snapshots for your ML notebooks – so you can back up, share, and rerun your work without “it worked yesterday” panic.

  • Never lose work again – capture timestamped capsules at real project milestones.
  • Share notebooks that actually run – send a clean package with code, deps, and data layout.
  • Understand your project months later – see what changed at each step.
Get early access See how it works

Built for students & researchers. Your data stays where it is – NoteCapsule runs inside your Jupyter / Colab environment.

Colab / Jupyter 
!pip install notebookcapsule -q

from notebookcapsule import create_capsule

create_capsule(
    name="resnet_baseline",
    data_dirs=["./data"]
)

📦 Capsule created:
./capsules/2025-11-23_resnet_baseline/

– notebook.ipynb
– requirements_suggested.txt
– data_manifest.json
– README_template.md

If your ML project lives only in a notebook, it’s one crash away from disappearing.

Right now, your entire Jupyter or Google Colab project is a single notebook plus “some data in Drive”. That’s fragile.

  • “It was working yesterday…” until the Colab runtime reset and you lost hours of work.
  • “Run all” behaves differently from running cells manually because of hidden state.
  • When you share your notebook, others hit ModuleNotFoundError and broken file paths.
  • Your folder is full of project_final, project_final2, project_final_really_final.ipynb.
  • You come back after a month and can’t remember what anything does.

Auto-save and “Download .ipynb” keep code, not context. When something breaks, you’re on your own.

Without NoteCapsule 
project.ipynb
project_final.ipynb
project_final2.ipynb
project_final2(backup).ipynb
data/ (somewhere in Drive)
env? (no idea)

Downloading the notebook sometimes is not the same as having a reusable, sharable project snapshot.

NoteCapsule turns your notebook into a reusable project asset.

Every time you hit a milestone, create_capsule(...) builds a Capsule – a self-contained snapshot with the notebook, suggested dependencies, data layout, and run instructions.

🛡 Protect your work

  • Create timestamped Capsules under ./capsules/ at real project checkpoints.
  • Copy your current notebook and metadata into a dedicated folder.
  • Stop relying only on Colab history or local autosave to rescue you.

🌐 Share notebooks that actually run

  • Auto-generate requirements_suggested.txt from your imports.
  • Record expected data files and layout in data_manifest.json.
  • Provide a README_template.md that explains how to run it from scratch.

🔁 Understand & maintain over time

  • See a simple Capsule history: initial_eda → baseline_cnn → resnet_augmented.
  • Roll back to a known-good Capsule if you break the notebook.
  • Help future-you, reviewers, and recruiters quickly understand your work.

How NoteCapsule works (it’s just one Python package)

No new platform to learn. Just a lightweight library you call from your existing Jupyter Notebook or Google Colab workflow.

Step 1

Install once in your notebook

!pip install notebookcapsule -q

Works in Google Colab, Jupyter Notebook, and VS Code notebooks – anywhere you can run pip.

Step 2

Create a Capsule at a milestone

from notebookcapsule import create_capsule

create_capsule(
    name="resnet_baseline",
    data_dirs=["./data"]
)

This creates ./capsules/<timestamp>_resnet_baseline/ with your notebook, suggested requirements, data manifest, README template, and metadata.

Step 3

See your Capsule history

from notebookcapsule import list_capsules

list_capsules()

Get a quick overview of all Capsules you’ve created for this project, instead of a folder full of random notebook backups.

Optional

Export & share as a zip

from notebookcapsule import export_capsule

export_capsule("resnet_baseline")
# → capsule_resnet_baseline_2025-11-23.zip

Zip a Capsule and share it via Drive or GitHub. Your collaborator gets the notebook + dependencies + data layout + README in one place.

Is NoteCapsule for you?

If you spend most of your project time inside Jupyter Notebook or Google Colab, yes.

🎓 Final-year & MSc students

ML / DS projects in Colab or Jupyter. Use Capsules as safe checkpoints for your thesis or capstone and as solid artifacts in your portfolio.

🧑‍🔬 Early-career researchers

Experiments and analysis in notebooks. Keep reproducible snapshots you can share with co-authors and rerun ahead of deadlines.

👩‍🏫 Instructors & mentors

Ask students to submit Capsules instead of raw notebooks. Spend less time debugging environments and file paths, and more time reviewing the work itself.

“Just download the notebook” vs creating a Capsule

Just download it
  • You get only the .ipynb file.
  • No list of dependencies or versions.
  • No record of which data files or folders were used.
  • No run instructions for other people.
  • Backups named final2_really_final.ipynb that all look the same.
Create a Capsule
  • Notebook plus suggested requirements_suggested.txt.
  • data_manifest.json with expected data paths and sizes.
  • README_template.md explaining how to run from scratch.
  • Clean, timestamped folders under ./capsules/.
  • Easy to zip, share, and restore later.

From the NoteCapsule blog

Deep dives on Jupyter Notebook, Google Colab, and reproducible ML projects – written to answer the exact problems people search for.

Google Colab · Backups

How to backup Google Colab notebooks (without losing work)

Colab timeouts and resets can wipe hours of training. Learn a simple workflow to keep your notebooks, outputs, and project state safe – and how NoteCapsule fits in.

Jupyter Notebook · Recovery

Jupyter Notebook backup & recovery: avoid “I lost my work” moments

What really happens when Jupyter crashes? We cover autosave, checkpoints, Git – and when you should add Capsules on top for project-level safety.

ML projects · Sharing

How to share Jupyter/Colab ML projects that actually run on other machines

Stop sending broken notebooks. This guide shows how to package code, environment, and data so that your project is truly reproducible for guides, teammates, and recruiters.

FAQ

Does NoteCapsule upload my notebook or data?

No. Early versions are designed to run in your environment. NoteCapsule reads your Jupyter / Colab notebook and directory structure to create a Capsule folder; your large dataset files stay wherever you already store them.

Do I need Git, Docker, or Conda to use this?

No. You just need Python and pip. If you already use Git, Capsules fit nicely into your repo – but they’re useful even without any extra tooling.

Is this a backup tool or a reproducibility tool?

Both. Each Capsule is a smarter backup: it includes your notebook plus the minimum reproducibility context (deps, data layout, README) so you can rerun and share your work with confidence.

Is NoteCapsule free?

During early access, yes. We plan to keep a generous free tier for students and solo researchers, and may add paid features later (cloud sync, share links, instructor dashboards).

Will NoteCapsule slow down my notebook?

Capsule creation is a quick file and metadata operation. You run it at milestones, not on every cell, so it shouldn’t affect your normal iteration loop.

Want NoteCapsule for your current ML project?

If your Jupyter or Google Colab notebook is where your project lives, NoteCapsule is your safety net. Join early access and we’ll send you a short guide, an example Capsule, and install instructions.