{"version":"2.1.282","anchor":"new-plugin-cache-slot-handedin-is-read-by-agents-commands-a","canonical_anchor":"new-plugin-cache-slot-handedin-is-read-by-agents-commands-a","heading":"Plugin commands, skills and agents can in principle be handed in instead of loaded","tier":"soon","area":"Plugins","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.282\/e\/new-plugin-cache-slot-handedin-is-read-by-agents-commands-a","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.282","markdown":"### Plugin commands, skills and agents can in principle be handed in instead of loaded\n\nLoaders for plugin commands, skills and agents now first check for content handed in from outside, but nothing in this build ever hands any in\n\n**Unclear.** The slot might still be filled in some way that was not found.\n\n**What**\n\nPlugins can add commands, skills (packaged instructions Claude can use) and agents (helper assistants Claude can hand work to). The code that loads these for a session now first checks whether they were already supplied from outside. If so it uses them instead of loading the plugins itself. Nothing in this build ever supplies them, so plugins are still loaded the usual way.\n\n**Why**\n\nThis looks like groundwork for letting a host program, such as a program built on Claude Code or a parent process, pass plugin content in directly instead of Claude Code reading it from disk.\n\n- Area: Plugins\n- Tier: Nothing to try yet\n- Useful: 2\/5\n- Signal: 3\/5\n- Present in the build but not switched on"}