Whenever I add a skill using npx skills add owner/repo Cursor initially list them under “Skills”, but after a few seconds they get moved into “Rules”.
I have a specific skills where I have added disable-model-innovation. The docs state:
By default, skills are automatically applied when the agent determines they are relevant. Set disable-model-invocation: true to make a skill behave like a traditional slash command, where it is only included in context when you explicitly type /skill-name in chat. Agent Skills | Cursor Docs
However, my skill is not available whenever I type slash /. Other skills are not available using slash neither.
Hey, thanks for the report. A couple of things to check:
After running npx skills add, where exactly do the files end up? Can you confirm the path? Is it .agents/skills/<skill-name>/SKILL.md or somewhere else?
Can you share the frontmatter from one of the affected SKILL.md files? Specifically the name, description, and disable-model-invocation fields. The name field has to match the parent folder name exactly. If it doesn’t, Cursor might not recognize it as a valid skill.
A screenshot of Cursor Settings > Rules showing how these show up after the reclassification would help.
According to the docs Agent Skills | Cursor Docs, skills should appear in the “Agent Decides” section under Rules, and /skill-name should work in chat. If they’re being moved to a different section, that usually means Cursor isn’t recognizing them. That can be caused by the file structure or how the frontmatter is being parsed.
So this is an easy mistake to make (I made it too).
Cursor does not automatically load skills from ~/.agents/skills. However, .agents/skills is supported for project-level skills. I think it would be less confusing if both the global skills and project-level skills use the same relative load paths. But for now, you can work around this by sym-linking ~/.agents/skills to ~/.cursor/skills.
