Open-science toolkit

Git & GitHub, without the headache

A free, beginner-friendly path into version control for the ManyBabies community — built for R-centric, OSF-organized, multi-lab work, with a real focus on analysis code, reproducibility, shared data, and QC. Every link was checked live and is free to use.

~50 vetted resources 3 skill tiers Verified June 2026 No cost · no fluff
Filter by level:

Start here by tier

Not sure where to begin? Find your level and do these three things first.

Tier 1 — Never touched a terminal
T1

You want to click, not type.

1. GitHub Hello World — 20 min in the browser, no install.
2. GitHub Desktop — commit & push by clicking.
3. Connect GitHub to your OSF project — code next to your data.

Tier 2 — New to Git
T2

You want the mental model + a clean workflow.

1. Happy Git with R — Git from inside RStudio.
2. A workshop — Software Carpentry or a free CodeRefinery session.
3. Learn Git Branching — makes branches finally click.

Tier 3 — Git keeps biting you
T3

You want best practices & recovery.

1. Dangit, Git!?! — copy-paste fixes for "I broke it."
2. Git Flight Rules + Pro Git ch. 7 — the deep manual.
3. Fix ugly notebook diffs — nbstripout & Quarto freeze.

1

Beginner on-ramps (GUI-first)

Get to your first commit and push without typing a single Git command.

GitHub · Interactive tutorial · Free · 20–30 min

Runs entirely in the browser — explicitly "no command line, no install." The lowest-friction first experience.

GitHub · Free desktop app · ~15 min to first commit

Point-and-click commit / push / pull. See the Getting Started docs. 100% GUI, open-source.

GitHub Docs · How-to · Free · 5 min

Edit a README, codebook, or script and commit it straight from the browser's pencil icon.

Atlassian · Illustrated explainer · Free · 15 min

Jargon-light "why version control exists" — good pre-reading to convince skeptics. No commands.

GitKraken · Video series · Free to watch · ~90 min

Visual, concept-first videos.

Heads up: the GitKraken app's free tier is public-repos-only — fine for learning, not for private lab repos.

2

Git for researchers & reproducibility

Git framed the way ManyBabies already thinks — provenance, reproducibility, open science.

Jenny Bryan · Free online book (CC BY-NC) · modular

The canonical resource for R users who aren't software devs; "Early GitHub Wins" + the RStudio Git pane get you committing from the IDE, no terminal.

Software Carpentry · Free lesson (CC BY) · ~half day

The gold-standard researcher Git lesson; includes an explicit Open Science episode.

Alan Turing Institute · Free handbook (CC BY) · 2–4 h

Why commit history is a scientific record; frames Git in reproducibility terms.

CodeRefinery / NeIC · Free lesson (CC BY)

Rewritten 2024 specifically for researchers, not engineers; companion git-collaborative lesson covers PRs & forks for multi-lab work.

Library Carpentry · Free lesson (CC BY) · ~half day

Gentle, non-coding entry — good for lab managers, RAs, and coordinators.

Lennart Wittkuhn (U Hamburg) · Free online book (CC BY)

The most recently maintained research-oriented Git book; covers GUI alternatives, with quizzes & exercises.

Poldrack Lab · Free handbook

Psychology-native language (preregistration, OSF, open data); T3 chapters reach GitHub Actions / containers.

Jenny Bryan · The American Statistician 72(1):20–27, 2018 · 30–45 min

The best "why bother" read for a stats / R audience.

Access: publisher is paywalled — the linked free PeerJ preprint (#3159) is the one to read.

Vuorre & Curley · AMPPS 2018 · 45–60 min

The only published Git tutorial by and for psychologists, using RStudio's GUI.

Access: publisher (SAGE) paywalled; the linked PsyArXiv preprint is free.

3

Free courses, webinars & continuing education

Structured, guided learning. There's no formal CEU for Git in psychology — Carpentries & CodeRefinery are the closest research-grade equivalents.

GitHub Education · Interactive, self-paced · Free · <1 h

Bot-guided exercises inside your own repo.

Heads up: if the skills.github.com portal shows a CSS error, open the exercise repo link directly — the exercise itself works.

CodeRefinery / NeIC · Free live online + open recordings

The strongest continuing-ed fit; research-focused. Next runs Sep 22–24 & Sep 29–Oct 1, 2026; recordings CC-licensed.

The Carpentries · Live workshops + free self-study · ~half day

Built for scientists; many universities host free editions, and a consortium member can host one.

Google · Coursera · MOOC · ~20 h

Well-structured video course with hands-on labs.

Access: free to AUDIT — videos free; graded work + certificate are paid. Click "Audit."

Udacity · Video course · Genuinely free · ~4–6 h

One of the few truly free (not just audit) structured courses.

Microsoft Learn · Interactive module · Free · ~1 h 45

Browser-only; the 58-min guided tour suits terminal-phobic T1.

freeCodeCamp · YouTube video · Free · 1 h 8

One-video crash course; easy to drop in a consortium Slack.

Codecademy · Interactive browser sandbox · ~2–4 h

No local install needed.

Access: freemium — core lessons free; certificate / AI / some projects need a paid plan.

Center for Open Science · Free live webinar + recordings · ~1 h/mo

Not Git per se, but covers OSF↔GitHub linking, preregistration, and data sharing — your ecosystem.

4

Notebooks & analysis-code version control

The fix for ugly, unreviewable diffs from .ipynb, .Rmd, and .qmd files.

Jenny Bryan · Ch. 12 · Free book chapter · 30–45 min

The canonical RStudio Git-pane setup; commit / push from the IDE, no terminal.

Jenny Bryan · Ch. 18 · Free chapter · 30 min

Use output: github_document so .Rmd diffs show readable Markdown, not HTML blobs.

Posit · Tool docs · Free · 20 min

freeze: auto version-controls outputs so collaborators don't re-run expensive code → stable, reviewable diffs.

Jupyter Project · Tool docs · Free (MIT)

Cell-level, content-aware diffs + a visual mergetool; auto-resolves execution-count conflicts. The fix for Jupyter merge hell.

Open source · Git filter / pre-commit hook · Free · 15 min

A one-time nbstripout --install keeps multi-MB outputs & images out of Git; optional Action enforces clean notebooks on PRs.

Open source · Tool docs · Free

Pair .ipynb with a .py/.md so you commit a diff-friendly script; supports the Quarto Markdown format.

ReviewNB · GitHub app · Free for public/academic repos

Rendered, cell-level notebook diffs + inline comments inside PRs.

Heads up: private repos capped (10 PR/mo) — but ManyBabies public repos get unlimited free use.

Toptal · Web tool · Free · 5 min

Generate a correct .gitignore for R, Python, JupyterNotebook, RStudio — stops accidental .RData / __pycache__ / .ipynb_checkpoints commits.

GitHub · Reference (CC0)

The authoritative R and Python templates GitHub pre-populates; customize (e.g., add _freeze/ for Quarto).

Google · Interactive · Free w/ Google login · 20–30 min

Open / commit notebooks to a GitHub repo from Colab, no local setup.

Heads up: needs a Google account; pair the repo with nbstripout (Colab won't fix diffs itself).

5

Shared datasets, data collection & QC workflows

When plain Git is the wrong tool for big/binary data — and how to run QC through GitHub.

Turing Institute · Free chapter · 20–30 min

Read this first: why plain Git fails for large/binary data, and Git LFS vs git-annex vs DataLad. The conceptual bridge.

GitHub · Tool · Free · 15 min

Version big files via pointers + .gitattributes. Good for modest files.

Heads up: GitHub's free LFS quota is ~1 GB storage/bandwidth — for large data prefer DataLad/GIN or OSF.

GitHub Docs · Reference · 10 min

Per-file ceilings (2 GB Free/Pro) and quotas — read before betting your data architecture on LFS.

DataLad community · Free book

The most thorough resource for distributed multi-lab data: nested datasets (one per site), provenance, on-demand retrieval, sensitive-data collaboration. Big files never enter the Git store.

DataLad · Free (BSD)

Standard in open neuroscience (OpenNeuro / NEMAR); publishes to GitHub, S3, Figshare. Pair with GIN (gin.g-node.org) — free academic storage.

Iterative · Tool docs · Free/OSS · 30–60 min

Lighter-weight alternative to DataLad for Python-comfortable labs; also versions ML pipelines; has a VS Code extension.

Center for Open Science · Workflow guide · Free · 15 min

Because ManyBabies lives on OSF: link the analysis repo to the OSF project in the browser so code sits beside preregistrations + data.

GitHub Docs · Reference/workflow · Free · 20–30 min

Turn GitHub into a data-QC tracker: an Issue per lab anomaly, a Projects board to track which sites passed QC.

GitHub Docs · Reference · Free · 10–15 min

Data QC as code review: gate incoming data/scripts with comment / approve / request-changes + CODEOWNERS auto-assignment — auditable quality control.

GitHub · Docs · Free for public repos · 30–60 min

Auto-run a validate_data.R on every PR and post a pass/fail check before a human looks. Scales QC across dozens of sites.

The consortium · Public org · all tiers

Your own best teacher: browse mb1-analysis-public & mb2-analysis (real collaborative R analysis), mb3-rules, the validator Shiny data-upload QC app, and mb1-cdi-followup. Real, domain-native models.

6

Cheat sheets, best practices & rescue

For the "I use Git and it keeps biting me" crowd — undo mistakes, fix conflicts, recover lost work.

Katie Sylor-Miller · Recovery cookbook · Free · skim 10–15 min

Scenario-first fixes anchored on git reflog. Use the clean-language dangitgit.com mirror in training materials. (original here.)

k88hudson · Recovery cookbook (CC BY-SA) · reference

The most comprehensive "things went wrong" manual (42k★, maintained 2026): bad merges, lost commits, accidental force-push, rebase disasters.

Chacon & Straub · Free book (CC)

The authoritative reference; Ch. 2.4 (Undoing Things) and Ch. 7 (Reset Demystified, Advanced Merging, rerere, Stashing) explain the why.

Atlassian · Tutorials · Free

Best diagrams for reset vs checkout vs revert, and a reflog deep-dive.

Peter Cottle · Interactive visualizer · Free/OSS · 1–4 h

Watch the commit graph move — the fastest way to truly understand branches, merges, and rebase (the root of most T3 pain).

GitHub Education · 2-page printable · Free

Pin it above your desk: setup, staging, branching, remotes, stash, reset.

Chris Beams · Article · Free · 15–20 min

The seven rules; clean history = a reproducibility record.

Community spec · Free · 15 min

feat/fix/docs… types map cleanly to data-cleaning, analysis updates, and pipeline changes; enables auto-changelogs.

GitHub Docs · Workflow guide · Free · 10 min

The branch → PR → review → merge loop multi-lab projects run on; bridges OSF-only users into PRs.

No resources match that search. Try a broader term, or reset the level filter to All.

ManyBabies-specific notes

  • Learn from your own repos. github.com/manybabies is public — point trainees at mb1-analysis-public, mb2-analysis, mb3-rules, the validator Shiny QC app, and mb1-cdi-followup. Domain-native examples beat generic dev tutorials.
  • OSF is the hub. Connect each analysis repo to its OSF project (browser, no terminal) so versioned code sits beside preregistrations and data.
  • A workable consortium workflow. Contributors branch → open a PR → a data coordinator reviews (data QC = code review) → merge. One Issue per lab anomaly, a Projects board to track which sites cleared QC, and a GitHub Action to auto-run validation on every PR.
  • Don't put raw infant data or large binaries in plain Git. Small files → Git LFS (mind the ~1 GB free cap). Large or many files → DataLad + GIN (free for academics) or keep data on OSF Storage and commit code + pointers.
  • Keep notebooks reviewable. nbstripout for Jupyter; output: github_document for .Rmd; freeze: auto for Quarto — otherwise PRs are unreadable.

Appendix — one-page everyday Git

The everyday loop

everyday.sh
git clone <url>          # get a repo (once)
git status               # what changed?
git pull                 # get teammates' latest BEFORE you start
git add <file>           # stage a change   (git add -A = everything)
git commit -m "message"   # save a snapshot
git push                 # share it

Branching — the safe way to work

branch.sh
git switch -c my-analysis    # create + move to a new branch
git switch main              # move back
# then open a Pull Request on GitHub to merge

"I broke something" — don't panic

rescue.sh
git restore <file>           # discard unstaged changes to a file
git restore --staged <file>  # unstage (keep the edits)
git reflog                   # find lost commits — your time machine

When in doubt: dangitgit.com or the printable cheat sheet. In R? Do all of this from the RStudio Git pane (Happy Git ch. 12).

Verification & access caveats

All resources were checked against their live source in June 2026 and are free to use. Most are confirmed live & free with no caveats — these carry one worth knowing.

ResourceStatusCaveat / note
GitHub SkillsConfirmed*Portal may throw a CSS error — open the exercise repo directly. Exercises work.
GitKraken (app)PartialFree tier is public-repos-only; private repos need a paid plan / student pack. Videos are free.
Google "Intro to Git/GitHub" (Coursera)ConfirmedFree to audit only — videos free; graded work + certificate are paid.
Codecademy Learn GitConfirmedFreemium — core lessons free; certificate / AI / some projects are paid.
ReviewNBConfirmedFree & unlimited for public/academic repos; private capped at 10 PR/mo. ManyBabies public repos = fine.
Google Colab ↔ GitHubPartialRequires a Google login; GPU/TPU is freemium. Integration confirmed via source repo.
Git LFS (on GitHub)ConfirmedTool free; GitHub storage/bandwidth capped at ~1 GB free — use DataLad/GIN or OSF for large data.
Bryan, "…Version Control?" (paper)PartialPublisher paywalled; the linked free PeerJ preprint (#3159) is the one to read.
Vuorre & Curley tutorial (paper)ConfirmedPublisher (SAGE) paywalled; the linked PsyArXiv preprint is free.
Connect GitHub to OSF (help article)PartialExact article restructured into the OSF Add-ons guide; workflow confirmed present.
Happy Git with RConfirmedLive & free; if a chapter link misbehaves, navigate from the table of contents.
Deliberately excluded / defunct (so you don't waste time):
  • GitHub Learning Lab → migrated to GitHub Skills; old URLs are dead.
  • rOpenSci Reproducibility Guide → archived/stale (no Quarto/renv).
  • PeerJ Preprints → closed to new submissions in 2019 (static archive; the Bryan preprint still loads).
  • Git-it desktop app → unmaintained since 2017.
  • LinkedIn Learning "Git Essential Training" → only conditionally free via a library login.
  • Coursera / DataCamp data-science Git courses → paid or audit-gated lab exercises.