The machine now carries its own manual for AI hands

Claude-Session: https://claude.ai/code/session_01Ms2FbjQy31TVf3CEvQhGM8

user/manifest

@@ -7,6 +7,10 @@
 #          repository. Use this for anything you want to update once and have
 #          change everywhere -- and for anything a tool writes into, because a
 #          skill installed on this machine should land in the checkout.
+#   linkdir  symlink each child of the source into the destination, which is
+#          created as a real directory. Use this where the destination holds
+#          more than this one source -- ~/.claude/skills also carries Panama's
+#          own shipped skills -- so it cannot be a symlink to any single place.
 #   copy   copy once, and never touch it again if the destination exists. Use
 #          this for files a machine legitimately diverges on.
 #
@@ -24,11 +28,15 @@
 link  agents/AGENTS.md   ~/.claude/CLAUDE.md
 link  agents/AGENTS.md   ~/.codex/AGENTS.md
 
-# Skills, at both paths that look for them. Linked as whole directories rather
-# than file by file, so a skill installed by any tool lands in the checkout and
-# `panama update` offers to commit it.
-link  agents/skills      ~/.agents/skills
-link  agents/skills      ~/.claude/skills
+# Skills, at both paths that look for them. Linked so that a skill installed by
+# any tool lands in the checkout and `panama update` offers to commit it.
+#
+# ~/.agents/skills is this directory and nothing else, so it stays one link.
+# ~/.claude/skills also holds the skills Panama itself ships, put there by the
+# link-skills stage, so it is a real directory with one link per skill -- and
+# this stage runs after that one, so a personal skill wins a name collision.
+link     agents/skills   ~/.agents/skills
+linkdir  agents/skills   ~/.claude/skills
 
 link  agents/rules       ~/.claude/rules