Decompose verification rules into atomic sub-tasks with cost-aware methods.
npx skills add https://github.com/memect/kc --skill task-decomposition-memect
Please help me install this Agent Skill. Skill: task-decomposition Source: https://github.com/memect/kc/tree/main/template/skills/zh/task-decomposition Command: npx skills add https://github.com/memect/kc --skill task-decomposition-memect
This Skill provides a structured approach to break down verification rules into independent, testable sub-tasks and assign the most cost-effective method to each, enabling transparent testing and scalable rule verification pipelines.
Outline a decomposition plan for a given rule, breaking it into atomic sub-tasks and assigning initial methods (regex, code, llm, or manual) to each sub-task.
High-intent search queries and answers about installing and using this skill.
Use MECE-driven decomposition to split complex verification rules into independent, non-overlapping atomic sub-tasks. This ensures complete coverage and enables transparent testing across your compliance pipelines.
Use a four-dimension decision matrix to allocate regex, python_calc, llm_tier1, llm_tier2, template, or manual_review to each sub-task. This cost-aware allocation ensures structured outputs suitable for workflow integration.
Yes, MECE decomposition supports linear, convergent, and fan-out pipeline topologies for diverse document-verification scenarios. It generates explicit structured outputs for each sub-task to guide skill-to-workflow integration.
Apply manual review when the four-dimension decision matrix determines that automated methods like regex, python_calc, or LLM tiers are not cost-effective or lack the required accuracy for a specific verification sub-task.
Yes, this method applies to rule extraction, plan design, and workflow integration across compliance and document-verification pipelines. It handles MECE decomposition and outputs structured data to guide downstream actions.
hassanzohdyDefine and document five string-format predicates for input validation and filtering.
dd-xiaozhiDecompose complex development tasks into parallel sub-agent operations and synthesize results.
lleontor705Detect project tech stack and generate a validated skill registry.
ScientiaCapitalFine-tune Llama, Mistral, Phi, or Gemma models with LoRA and 4-bit quantization.
