What problem does it solve? When developers investigate how a repository works, their findings usually vanish after the conversation ends, forcing the same code-reading effort to be repeated later. This Skill turns the "ask question → read code → reach conclusion" loop into archived, searchable exploration documents with file:line evidence. ## Core Features & Use Cases - Three exploration types: question (answer a specific code question), module-overview (map a module's structure, entry points, and dependencies), and spike (lightweight probing of multiple technical directions without deciding). - Evidence-first documentation: every conclusion must trace back to 3-8 pieces of evidence annotated with file:line, with a conclusion-first quick-answer section and Mermaid diagrams for multi-module flows. - Lifecycle management: detects overlapping prior explorations, supports updating, superseding, and marking outdated documents, and archives results to codestable/compound/ with searchable YAML frontmatter. - Use Case: A new team member asks "how does the serial data parsing work in this repo?" The Skill reads the actual code, produces a dated explore document with a quick answer, cited evidence, and a Mermaid call-chain diagram, then archives it so the next person finds it in seconds. ## Quick Start Ask the AI to explore how a specific feature is implemented in this repository and archive the findings as an explore document.
