Intern Workbench is an open-source, GitHub-native framework for running a structured web-development internship. A company forks the repository, selects and customizes a curriculum, assigns tasks as GitHub Issues, and reviews the intern's work through pull requests. There is no dashboard, hosted backend, learner database, or AI detector: GitHub is the platform.
This is a structured internship framework, not a replacement for real engineering experience.
It removes repetitive procedural mentoring—where to work, how to branch, how to test, what a PR needs—so mentors can focus on reasoning, tradeoffs, code quality, and growth. The curriculum contains 206 consolidated core tasks from absolute beginner through junior-developer-ready practice.
FORK → CUSTOMIZE → INVITE INTERN → ASSIGN ISSUE → IMPLEMENT → TEST
→ PR → CI → REVIEW → FIX → APPROVE → MERGE → NEXT TASK
- Companies and mentors: MENTOR-GUIDE.md, then CUSTOMIZATION.md.
- Interns: ONBOARDING.md. The first successful PR is designed to fit in the first working session.
- Curriculum planners: CURRICULUM.md and schedules/README.md.
- Contributors: CONTRIBUTING.md and TASK-AUTHORING-GUIDE.md.
Permanent phase IDs such as JS-006 are independent from GitHub Issue and PR numbers. Guidance decreases from GUIDED to AMBIGUOUS; tasks move from explicit implementation to investigation and product delivery. Shared workflow documentation avoids repeating a textbook inside every task.
AI assistance is allowed and must be disclosed. Evaluation is based on verification and demonstrated understanding—not unreliable authorship detection. See AI-USAGE.md.
npm run validateThe validator checks metadata, IDs, prerequisites, links, required task sections, policy placeholders, and repository structure. CI runs the same objective checks; mentors retain responsibility for subjective judgment.