Evaluates coding tasks and assigns weighted complexity scores to ensure manageable, atomic work units.
Complexity Scorer is a specialized analysis engine designed for the Task Master plugin that prevents over-scoped coding tasks by enforcing technical rigor. By evaluating tasks across eight critical dimensions—including file impact, testing requirements, and architectural risk—it identifies when a task is too complex to be handled effectively in a single pass. It enforces a strict 'atomic task' ceiling of 4, automatically flagging complex work for decomposition into smaller, more reliable sub-tasks, which ensures Claude operates within high-confidence parameters and reduces the likelihood of regressions.
Key Features
010 GitHub stars
02Batch mode support for scoring multiple tasks simultaneously during planning phases
03Automatic 'splitRequired' flagging for tasks exceeding the atomic threshold
04Weighted 8-factor scoring system including files affected, risk, and testing needs
05Detailed JSON output with factor-by-factor scoring and written justifications
06Context-aware evaluation based on existing codebase patterns and dependencies
Use Cases
01Pre-flight analysis of feature requests to ensure they are properly decomposed before execution
02Improving AI reliability by preventing Claude from attempting 'God-tasks' that touch too many files
03Identifying high-risk architectural changes that require more intensive testing or review
