jobs-to-be-done-extractor

# Jobs-to-be-Done Extractor This skill reads a folder of custdev transcripts and interview notes (md or txt files) and extracts structured Jobs-to-be-Done (JTBD) statements with evidence quotes and cross-file frequency counts. Output is a prioritized JTBD map with confidence labels ready for roadmap and prioritization work. **Input:** - Folder with custdev files (.md or .txt) — transcripts, interview notes, open-ended survey responses **Output:** - Markdown JTBD map with ranked statements, evidence quotes, source references, and a Patterns & Gaps section --- ## Language Detection Detect the user's language from their message: - If Russian (or contains Cyrillic): respond in Russian - If English (or other Latin-script language): respond in English - If ambiguous: respond in the language of the trigger phrase used --- ## Instructions ### Step 1: Validate and Read Input Files 1. Ask the user to point to the folder containing custdev files (if not already provided). 2. List all .md and .txt files in the folder. - If no files found: stop. Report: "No input files found. Point me to a folder with custdev transcripts or notes (md or txt)." - If files found: report count — "Reading N files from [folder]." 3. Read each file sequentially. - If a file is empty or unreadable: skip it, note filename — "Skipped: [filename] — empty or unreadable." Continue with remaining files. 4. If only 1 file found: note — "Single source — frequency analysis not applicable. All statement...