Prompt
Design or improve skills to be reused based on the following data. [input] AI and execution environment to use: Codex desktop app, skills used in local projects Existing skill files and references: SKILL.md contains multiple task modes and lengthy examples in one file. Conditions to maintain and scope of work: Written in Korean. Existing files are preserved. Provides registration files and verification methods without assuming tools that do not exist. [Tasks to be solved in this skill] Reduce duplication and generalities by looking at what kind of judgment each guideline is needed for. Move common procedures to SKILL.md and long descriptions of specific situations to a reference file that is read when actually needed. Do not create unnecessary folder structures for short skills. Preserve call conditions, required constraints, existing input and output, and check links before and after file movement. [Output for registration] Check the skill format supported by the usage environment and write the entire SKILL.md that can be saved immediately. The name·description should reveal the function and point of use, and the body should include work flow, input, output, exception handling, and completion criteria. Add reference materials, scripts, and templates only when actually needed, and distinguish the path and contents of each file. The registration location and calling method will be guided based on confirmed environmental information. If you can't create or test the file yourself, provide the content and mark it as not running. Do not create tools/paths/installations that you have not confirmed. If important information is missing, only ask questions that will change the outcome first. [Case to be verified] Provide example requests and expected behavior, including reproducing existing representative work, missing reference files, and unused documentation. Distinguish between the actually performed verification and the proposed verification, and if there is a failure, let me know what needs to be corrected.