Inline code with test cases
Coding questions support starter code, sample tests, hidden checks, run permissions, and execution history. Students write and submit code directly.
CODING & PROJECTS
Assess the reasoning behind technical work
AI-generated code can pass automated tests and surface-level review. Pruuva helps instructors evaluate whether students understand their implementation, design decisions, edge cases, and project structure.
Repository-based submissions
Git project submissions analyze repository structure, commit history, required files, and implementation evidence from GitHub, GitLab, or Bitbucket.
Technical defense
Follow-up questions target logic flow, architecture, constraints, trade-offs, debugging choices, and how the student would adapt their solution.
Why code submissions need capability evidence
A student submits working code. Automated tests pass. The output is correct. But did the student write it? Do they understand why it works? Could they modify it if a requirement changed?
These are the questions code review cannot answer at scale. Pruuva generates follow-up questions from the student's specific code, targeting the decisions, patterns, and edge cases in their implementation.
The result is structured evidence showing whether the student can explain their own code, not just produce a correct output.
HOW IT FITS
Built around the evidence instructors need to trust a decision.
01 / Configure
Set the exercise or project requirements
Define language, starter files, test visibility, resource limits, repository expectations, and the rubric dimensions you care about.
02 / Execute
Collect outputs and execution evidence
Run code checks, store results, and separate public feedback from hidden grading evidence. Repository submissions are analyzed for structure, history, and implementation.
03 / Defend
Probe understanding beyond passing tests
Follow-up questions target specific code paths, commits, edge cases, and design decisions. Students explain the reasoning behind their implementation.
ACADEMIC INTEGRITY FOR CODE
AI-generated code is a real challenge. Capability evidence is a practical response.
Detection does not work for code
AI code detectors are unreliable. Code has structure, patterns, and conventions that make statistical authorship detection even less meaningful than for prose.
Tests prove correctness, not understanding
Passing automated tests shows the code works. It does not show the student understands why it works, how to modify it, or what trade-offs they considered.
Code review does not scale
Manually reviewing each student's code for signs of understanding works in small classes. At scale, instructors need a structured follow-up process.
Oral defense, automated
Pruuva provides the same pedagogical value as a live code walkthrough, but runs asynchronously for every student. The instructor reviews the evidence report, not a live recording.
USE CASES
Technical assessment in practice
Introductory programming
Verify that students understand loops, conditionals, and data structures in their own code. Follow-ups ask them to trace execution, predict outputs, and explain their approach.
Data structures and algorithms
Beyond correct implementations, probe for understanding of time complexity, space trade-offs, and why the student chose a specific approach over alternatives.
Software engineering projects
Analyze repository structure, commit history, and architecture decisions. Follow-ups target design patterns, API choices, and how the student would handle new requirements.
Team projects
Each team member gets their own follow-up. Questions target their specific contributions, drawing from commit history and the portions of the codebase they worked on.
Capstone and thesis work
Deep-dive into methodology, architecture, and implementation decisions before committee review. Surface gaps early and give students a chance to demonstrate their understanding.
Lab and data analysis
Students submit their analysis code and results. Follow-ups probe for understanding of the methodology, data processing steps, and interpretation of results.
USE THIS WHEN
The question is not only whether work was submitted.
Use Pruuva when a grade, appeal, pilot, or academic-integrity review needs evidence that the learner can explain the submitted work in context.
The evidence needs to be reviewable later.
Pruuva keeps the artifact, follow-up, rubric findings, and instructor decision together so educators can revisit why a decision was made.
COMMON QUESTIONS
Frequently asked questions
What programming languages are supported?
Python, JavaScript, TypeScript, Java, C++, C, Go, Rust, and more. The follow-up questions are language-aware and target the patterns and idioms of the student's chosen language.
Can I use Pruuva with auto-grading tools?
Yes. Pruuva adds the understanding layer that auto-graders cannot provide. Use auto-grading for correctness and Pruuva for capability evidence.
How does Pruuva handle repository submissions?
Students connect their GitHub, GitLab, or Bitbucket repository. Pruuva analyzes structure, commit history, and code content. Follow-ups can target specific commits, files, or architectural decisions.
What about students who pair-programmed?
Each student gets their own follow-up, even on shared repositories. Questions target their individual understanding and specific contributions to the project.
RELATED PRUUVA RESOURCES