{"version":"2.1.323","anchor":"the-skill-tool-description-was-rewritten","canonical_anchor":"the-skill-tool-description-was-rewritten","heading":"The instructions Claude reads about using skills were reworded","tier":"internal","area":"Skills","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323\/e\/the-skill-tool-description-was-rewritten","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323","markdown":"### The instructions Claude reads about using skills were reworded\n\nClaude Code reworded the Skill tool's instructions about calling a matching skill first and following its steps\n\n**What**\n\nA skill is a packaged set of instructions that you or your project set up for a particular kind of task, such as deploy steps or a review checklist. Claude uses the Skill tool to load one. The description of that tool, which is the text Claude reads about how to use it, was reworded in this release.\n\nIt now tells Claude that when a listed skill covers the task at hand, it should call the tool first. The skill's instructions then load into the turn for Claude to follow in place of its usual approach. Some skills instead run in a subagent, a separate helper copy of Claude, and return the finished result.\n\n**Why**\n\nChanging Claude's instructions changes how it behaves. This text decides when Claude reaches for your skills and what it does with them once loaded.\n\n- Area: Skills\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 0\/5\n- Scope: individual\n- Heads-up: no"}