What problem does it solve? Analyzing large code files, data exports, or reports directly can flood the AI's context window and degrade answer quality. This Skill prevents that by planning the analysis approach upfront — deciding which files to read directly, which to read partially, and which to delegate to subagents — before any file content is loaded. ## Core Features & Use Cases - Pre-read inventory: Collects file existence, line counts, and types (code, data, report, document) without reading content, producing a summary table. - Decision matrix: Routes files by size and type — small files read directly, large .bsl modules delegated to a 1C code explorer subagent, XML/HTML/CSV data files delegated for structure extraction. - Subagent question templates: Provides ready-made prompt templates for algorithm analysis, call/dependency search, XML/JSON structure extraction, and HTML report parsing. - Synthesis protocol: Cross-analyzes subagent results, fills gaps with targeted fragment reads, and composes the final answer. - Use Case: You need to understand how a 2500-line 1C object module maps records from an XML export shown in an HTML report. The Skill inventories all three files, dispatches focused subagents in parallel, then synthesizes their structured summaries into one answer. ## Quick Start Ask the assistant to analyze the attached large module and data files, and it will first present a brief analysis plan before reading anything.
