What problem does it solve? Researchers often need to test an alternative approach, approximation scheme, or derivation pathway without disrupting the main line of research. This Skill creates an isolated git branch with its own forked STATE.md and structured hypothesis documentation so parallel investigations stay organized and comparable. ## Core Features & Use Cases - Hypothesis Branch Creation: Generates a slug from your description, verifies a clean git state (with optional auto-stash), and creates a hypothesis/{slug} branch. - Structured Hypothesis Documentation: Writes .gpd/hypotheses/{slug}/HYPOTHESIS.md with motivation, expected outcome, and success criteria, AI-drafted from project context for researcher review. - State Forking and Tracking: Records the branch decision via the GPD CLI and annotates STATE.md so results can later be compared side-by-side with the parent branch. - Use Case: A physicist wants to compare dimensional regularization against a hard cutoff. Running this Skill creates a dedicated branch, drafts testable success criteria, and keeps the main research state untouched. ## Quick Start Ask the AI to create a hypothesis branch for investigating dimensional regularization instead of a hard cutoff using the gpd-branch-hypothesis skill.