Capture Hidden Expertise from a Screen-Recorded Demonstration

ROLE You are an expert in eliciting procedural knowledge from demonstrations. OBJECTIVE Convert a recorded task demonstration into learning material for my AI twin, emphasizing hidden judgments rather than merely listing clicks. CONTEXT The demonstrated task is [TASK NAME]. The transcript or annotated recording is [DEMONSTRATION MATERIAL]. The intended learner is [INTENDED LEARNER]. INSTRUCTIONS Separate directly observable actions from inferred intentions. Identify moments where I pause, compare options, inspect a signal, undo an action, or skip an apparently available step. For each moment, draft a targeted follow-up question that would uncover the missing judgment. Produce an annotated sequence with the observed cue, action, likely rationale marked as provisional, and evidence location. Create a short interview agenda prioritized by which answers would most improve the twin's performance. Include environmental dependencies such as interface versions or required permissions. CONSTRAINTS Do not claim to inspect video unless the current tool supports the supplied format. If necessary, request a transcript or timestamped screenshots. Avoid fabricated statistics and flag assumptions. Do not treat cursor movement alone as proof of intention. OUTPUT FORMAT Observation table, hidden-knowledge questions, interview agenda, and a list of unresolved dependencies.