Orchestrates a multi-stage subagent pipeline to analyze large datasets or codebases that exceed standard context windows.
This skill enables Claude to process massive volumes of text, code, or data by distributing the workload across a structured hierarchy of subagents. It implements a two-stage fan-out pattern where 'Worker' agents analyze segments, 'Critic' agents review findings for consistency and gaps, and a 'Summarizer' synthesizes the final report. With built-in task tracking, failure recovery, and configurable effort levels, it is an ideal solution for comprehensive security audits, deep documentation reviews, or large-scale codebase analysis where accuracy and thoroughness are paramount.
Key Features
01Redundant review cycles (Critics) to identify gaps and inconsistencies
02Dynamic task dependency tracking with failure recovery and auto-splitting
035 GitHub stars
04Parallel subagent orchestration for Workers, Critics, and Summarizers
05Automatic corpus segmentation and token-limit management
06Configurable effort levels to balance analysis speed and depth
Use Cases
01Synthesizing insights from extensive legal, medical, or technical documentation sets
02Performing comprehensive architectural or security audits on massive codebases
03Identifying cross-module inconsistencies and patterns in large-scale data repositories