What problem does it solve? Summaries compress content but lose the argumentative structure, hedges, and non-obvious implications that make long-form material valuable. This Skill decomposes articles, papers, transcripts, and reports into their core thesis, argument chain, evidence-backed claims, and tensions — with every claim anchored to the source so you can verify it. ## Core Features & Use Cases - Structural Decomposition: Extracts core thesis, argument chain (linear or threaded), key claims with strength ratings, emphasis signals, facts, and tensions — not a summary. - Fidelity Safeguards: Runs a 10-check fidelity gauntlet covering hedge preservation, proper-noun audits, speaker attribution, inference labeling, and line-number anchoring to prevent hallucinated or misattributed content. - Content-Type Adaptations: Adjusts extraction strategy for transcripts (filler stripping, speaker attribution, STT garble correction), papers, opinion pieces, meetings, and reports. - Use Case: You have a 3,000-line podcast transcript. Run the skill to get a ~300-line extraction with the guest's thesis, threaded argument chain, memorable quotes with line references, and the tensions where the speaker's real thinking shows. ## Quick Start Ask the assistant to extract insights from a file, URL, or pasted text, for example: "Extract insights from this transcript at full depth."
