Documentation
Guides for apprentices, mentors, and curriculum content creators — apprenti.dev is Git-native, so these are guides to files and workflows, not a hosted product manual.
For Apprentices
Installing the app
Sideload Windows, macOS, Linux, or Android from GitHub Releases.
Getting started
Open or clone a curriculum repository and start working offline.
Working through your path
Today, Path, task work, notes, and time logs.
Submitting evidence for review
Build a submission your mentor can inspect, then track revisions.
Using AI assistance
What the AI tab and task assistants can and cannot do, and where your prompts go.
The five AI assistance modes
What explain, hint, quiz, coach, and solution each actually do.
Keeping your fork private
GitHub keeps forks of public repos public; GitLab lets you choose. What to do on each.
Search across your curriculum
Find any task, resource, or note instantly, fully offline.
For Mentors
Installing the app
Same sideloaded builds apprentices use — no store listing.
Reviewing submissions
Work the review inbox: approve, request revision, or discuss.
Mentoring & communication
Threads, feedback, and help requests, with no server in the middle.
Setting up a curriculum repository
Clone from GitHub or GitLab, and understand what stays private.
Tracking competency growth
Turn approved evidence into a defensible record of what an apprentice can do.
For Content Creators
Installing apprenti creator
Sideload the VSIX into VS Code, Cursor, or VSCodium — not a second app.
Curriculum structure basics
A curriculum is files: JSON metadata and Markdown instructions, versioned in Git.
Authoring with apprenti creator
The VS Code / Cursor extension for building a path visually.
Setting the AI policy for a task
Decide what AI may do on each task, and when disclosure is required.
How competencies and evidence connect
Which task fields point at which files, and how they're meant to be used together.
Designing competencies and levels
Backward design, writing levels that mean something, and avoiding competency sprawl.
Writing tasks that produce real evidence
Checkable acceptance criteria, AI-resistant evidence, and calibrating difficulty.
Locale overlays
Ship a curriculum in more than one language without duplicating content.
Publishing & forking a curriculum
Git remotes, upstream curricula, and sharing a fork safely.
Using AI skill packages
Install guides for Claude Code, Cursor, OpenAI Codex, OpenCode, Copilot, and a ChatGPT Custom GPT.
Hosting local AI models
Google Drive hosting, the catalog-building script, and which models fit a phone versus a desktop.