Refine code, text, or data iteratively against measurable criteria.
npx skills add https://github.com/witooh/skills --skill improve-witooh
Please help me install this Agent Skill. Skill: improve Source: https://github.com/witooh/skills/tree/main/skills/improve Command: npx skills add https://github.com/witooh/skills --skill improve-witooh
This Skill addresses the challenge of refining outputs (code, text, data) to meet specific, measurable standards, ensuring that improvements are objective and user-satisfactory.
Use the improve skill to refine the provided code snippet until it meets the specified performance benchmarks.
High-intent search queries and answers about installing and using this skill.
Objective output refinement scores artifacts against user-defined criteria at each iteration, tracking measurable improvements until the result meets the specified standards. It supports both single-dimension quick fixes and complex multi-dimensional tasks.
Improve code quality by providing a code snippet and specifying performance benchmarks; the skill refines the code iteratively while scoring it against those benchmarks. This ensures objective, measurable enhancement rather than subjective changes.
Yes, the iterative refinement process offers a Quick Mode for single-dimension improvements and a Full Mode for complex, multi-dimensional tasks. You select the mode based on the scope of the artifact refinement needed.
Yes, you can use content editing to make a technical document clearer and more concise by defining those attributes as criteria. The skill revises the document iteratively, scoring clarity and conciseness until you are satisfied.
You should avoid scored evaluation loops for optimization when an output lacks clear, measurable criteria for scoring. Without definable metrics to track progress at each step, the iterative refinement process cannot function objectively.
Structured discovery requires clarification to ensure the iterative refinement hits exact user requirements. The skill uses interactive questions to define the measurable criteria needed for objective scoring and progress tracking.
InverterNetworkEvolve prompts, regex patterns, SQL queries, and code snippets via LLM-driven search.
modu-aiGenerate API documentation from source code with multi-format outputs.
barnabasJDesign language-agnostic test strategies using TDD, BDD, and AAA patterns.
disciplin-run-orgExecute automated QA tests on code repositories and generate detailed reports.
