Unclear It is not clear whether the always-allow option actually appears for MCP skills, since prompts for them passed on remotely still leave it out.
MCP (Model Context Protocol) servers are outside programs that give Claude extra tools and skills. Some permission prompts arise for a reason that Claude Code's automatic safety check cannot approve. For those prompts, Claude Code used to always hide the option to always allow the action. It no longer does this when the request comes from an MCP skill.
A consent disclosure is a notice that has to be shown to you before you agree to something. When Claude Code cannot display one, it refuses the action. That refusal now has a separate version for MCP skills.
Permission prompts for MCP skills may now offer an always-allow option that was hidden before.
It is not clear whether the always-allow option actually appears for MCP skills, since prompts for them passed on remotely still leave it…