What problem does it solve? Conducting deep research on dozens of items one by one is slow and inconsistent. This Skill reads a research outline file and dispatches independent background agents for each item, producing uniform structured JSON results with resume support. ## Core Features & Use Cases - Outline-Driven Batch Research: Automatically locates an outline.yaml file, reads the item list and execution config, and processes items in configurable batches with user approval between batches. - Parallel Agent Execution with Resume: Launches background web-search agents that each handle multiple items, skipping items whose JSON output already exists so interrupted runs can resume. - Validated Structured Output: Each agent writes JSON conforming to a fields.yaml definition, marks uncertain values, and runs a validation script to confirm complete field coverage. - Use Case: You are compiling a dataset of 50 AI coding tools. Define them in outline.yaml with a shared fields.yaml schema, then run the skill to research every tool in parallel batches and collect validated Japanese-language JSON profiles in the results directory. ## Quick Start Run /research-deep in a directory containing an outline.yaml file to launch batched deep-research agents for every item in the outline.
