Decomposes complex codebase questions into parallel sub-agent tasks to generate cited research reports.
This skill automates the process of understanding unfamiliar or complex codebases by dispatching specialized sub-agents—Locator, Analyzer, and Pattern-Finder—to investigate specific aspects of a project simultaneously. It systematically breaks down broad research queries, performs deep-dive searches with intelligent synonym expansion, and synthesizes findings into a persistent Markdown report complete with mandatory file and line-number citations. Whether you are onboarding to a new project or performing an architectural audit, this skill ensures no technical detail is missed by leveraging parallel processing and rigorous evidence-based reporting.
主要功能
01Mandatory file:line citations for every technical claim made in reports
02Parallel sub-agent dispatching using specialized models for search and analysis
030 GitHub stars
04Automated question decomposition into independently answerable sub-tasks
05Persistent Markdown reporting with automated progress saving via the 2-Action Rule
06Intelligent synonym expansion to catch aliased concepts (e.g., auth vs. login)
使用场景
01Performing comprehensive security or authentication audits across large-scale projects
02Accelerating onboarding by mapping out core architecture and data flows in a new repository
03Identifying and comparing existing implementation patterns to ensure architectural consistency
