BESTSKILLSFOR CLAUDE
Head to head

Six pairs of skills that look like the same skill

A review tells you whether one skill is worth installing. It cannot tell you which of two to keep when both claim the same job. These pages take the six pairs where that question actually comes up and answer it.

  1. tdd or test-driven-development: which one should you install?

    Install one of them, not both. They describe the same discipline from two publishers, so keeping both leaves two overlapping descriptions competing for the same task. Pick the one whose house style matches how your team already writes tests.

  2. systematic-debugging or diagnosing-bugs for finding the cause?

    Both stop the agent from guessing at fixes, but they enter the problem from different ends. systematic-debugging holds a process until the cause is proven. diagnosing-bugs is closer to a working investigation of the failure in front of you.

  3. skill-creator or mcp-builder: which one are you actually building?

    They build different things. skill-creator helps you write a folder of instructions the agent reads. mcp-builder helps you write a server that exposes new tools. If the agent already has the access and only lacks the procedure, you want skill-creator.

  4. docx or pdf: which document skill do you need?

    The answer is decided by the file, not by preference. docx is for Word documents someone will keep editing. pdf is for a fixed document you need to read, split, fill or produce as a final artefact. Most document work ends up needing both.

  5. frontend-design or canvas-design: which one fits the work?

    frontend-design is for interfaces that will ship as code, with the taste applied to components, layout and states. canvas-design is for composed visual work on a canvas. If the output ends up in a repository, you want frontend-design.

  6. subagent-driven-development or superpowers: how much structure do you want?

    superpowers is a broad working method for a whole session. subagent-driven-development is the narrower question of splitting work across agents. One changes how every task is approached, the other changes how one large task is divided.