Code is text that tells a computer what to do. Git records versions of project files on your computer; GitHub stores a shared copy and the pages used to review proposed changes. This track teaches that workflow before adding Python or an AI coding tool.
Complete this track if you create or modify code, scripts, notebooks, LaTeX, or documentation in a GitHub repository.
This page summarizes the track. Follow the current mission in your local Passport; do not use the links below as a second checklist.
Prerequisites#
- IT and Research Safety Core
- Access to the relevant GitHub organization and training repository
Current Training Evidence#
Git checks the prepared identity and the bounded manual draft-PR exercise. Python and AI are separate assigned responsibilities; project access is separate from public training.
Use the current Passport lesson for sequencing. The competencies and evidence below describe later project capability and private owner review; they are not extra public submissions or automatic access grants.
Competencies#
You can:
- authenticate without exposing credentials;
- clone a repository and inspect its instructions;
- create a branch and keep changes focused;
- distinguish source, generated files, data, and secrets;
- run the repository's tests and quality checks;
- write a Conventional Commit such as
fix(parser): handle empty input; - open a pull request that reports only verification actually performed;
- recover from ordinary conflicts without destructive Git operations.
If your assigned route includes AI agents, you can also distinguish the model, provider/gateway, agent harness, and interface; constrain files, tools, data, side effects, and spending; and review AI-generated diffs.
Mission Coverage#
- Read Git and GitHub workflow.
- Read Python setup if the project uses Python.
- Complete First safe pull request manually.
- Complete Reproducible Python environment when relevant.
- Complete Project handover when you own a deliverable that another person must reproduce.
- If the assigned route includes AI agents, read Agents and interfaces, configure one documented access route, then complete Manual versus agent.
Fast Path#
Experienced contributors may move quickly through explanations, but still complete the small Git practice PR and the same automatic checks.
Evidence#
- A focused pull request from a non-
mainbranch. - At least one appropriate test or verification command.
- A Conventional Commit.
- A PR description with purpose, changes, affected interfaces/users, verification, risk, and AI assistance.
- A manual/agent comparison naming personally run checks and one accepted and one rejected or revised suggestion.
- No secrets, large data, or accidental generated files in the diff.
Understand Before Accepting AI Output#
- I can explain every changed behavior.
- I know which test supports the change.
- I checked the diff for secrets and unrelated files.
- I did not let the agent invent verification results.
- I know which service received context and which account would be billed.
Pass Criteria#
The trusted automatic checks pass, and the contributor can explain the change, the staged diff, and the verification they personally ran. Normal project PRs still follow that project's own human review policy.