What problem does it solve? Ad-hoc web searches produce shallow answers with one or two sources and no audit trail, making it impossible to verify claims or trace them back to real sources. This Skill runs a structured L2 research pipeline that scrapes full source content, detects contradictions, and grades confidence per claim. ## Core Features & Use Cases - Composed research pipeline: Builds on the L1 /research skill, then adds a reflection loop with gap analysis, two follow-up sub-questions, and 10-15 additional scraped sources. - Contradiction detection and confidence grading: Explicitly surfaces disagreements between sources and labels every major claim as High, Medium, or Low confidence. - Verification checkpoints: Four mandatory Bash checkpoints verify that searches, scrapes, and summaries actually exist on disk, preventing hollow synthesis from search snippets. - Use Case: When deciding whether to migrate from one technology to another, run this Skill to get a ~2000-word report synthesizing 20-30 sources, with a dedicated contradictions section and per-claim confidence levels. ## Quick Start Ask the AI to run deep research on your question, for example: use deep research to compare PostgreSQL and MySQL for a high-write analytics workload.
