Real Research is an AI-powered research workflow that runs inside VS Code. It pairs students with a Socratic AI tutor that walks them through every stage of a real scientific project — from question to presentation.
Explore the workflowReal Research combines structured pedagogy with AI mentoring to guide students through the full lifecycle of a scientific project.
Claude doesn't write your paper — it asks the questions your advisor would. Every stage uses Socratic probes to push your thinking and catch weak logic before it hardens.
From project setup to final presentation, every stage has substages, checklists, and quality gates. You always know what's done and what's next.
Causal, descriptive, or normative — each mode reshapes every stage with method-specific probes, expectations, and validation criteria tailored to your question type.
A library of method-specific agents — from regression analysis to case studies, data acquisition to causal inference — automatically assigned based on your research design.
Runs as a VS Code extension with a dedicated sidebar, file explorer, and Claude integration. Everything lives where you already write code and text.
All pedagogical content — agents, skills, stage contracts — is encrypted with AES-256 and decrypted only at runtime. Students interact through guided handoffs, never raw prompts.
Each stage has substages, quality gates, Socratic probes, and deliverables. Click a stage to explore what students work through.
The research mode you choose reshapes every stage. Probes, expectations, validation criteria, and even the specialist agents adapt to your question type.
For questions about causes, effects, and mechanisms. Stages emphasize identification strategies, counterfactuals, threats to internal validity, and tests of alternative explanations.
For questions about patterns, distributions, and taxonomies. Stages emphasize conceptual validity, measurement quality, comparison structures, and systematic classification.
For questions about values, policy choices, and institutional design. Stages emphasize evaluative criteria, tradeoff analysis, stakeholder perspectives, and principled comparison of alternatives.
Real Research orchestrates a structured loop between you and your AI tutor. Each stage follows the same pattern.
Click the stage in the sidebar. The server assembles a handoff pack with your prior work, stage instructions, and method-specific probes.
Claude reads the handoff and engages you in Socratic dialogue. It asks hard questions, pushes your reasoning, and flags gaps — but never writes your argument for you.
Based on the conversation, you produce the stage's deliverables: a framing document, analytical framework, evidence map, or manuscript draft.
The extension detects your result, runs it through quality checks, and advances you to the next stage. Your progress is tracked and versioned.