Claude Skill Best Practices

In this instructional video, Robyn Anderson warns that users often overload their AI assistants with many custom "skills" (keyword-triggered functionalities). She explains that AI models, such as Anthropic's Claude, may start ignoring or misusing older or rarely used skills, especially if they were added more than 30 days ago, leading to wasted tokens and sub‑optimal performance. Anderson cites Anthropic’s guidance that excessive skill clutter can cause the model to ignore user input. To address this, she offers a prompt that audits a user's entire skill setup, removes unused or outdated skills, and reorganizes the remaining ones so the AI can better understand the user's intent and suggest more efficient workflows. She emphasizes that regular, weekly skill pruning keeps the AI responsive, future‑proofs workflows, and reduces token consumption. The video ends with a call‑to‑action: viewers should save the video, comment "jam," and she will provide the full audit prompt.










