Claude Code skill allowed-tools: a one-turn grant, not a sandbox

You add allowed-tools: Bash(git add *) Bash(git commit *) to a skill. You type /commit , and Claude stages and commits without asking. You send one more message, ask for another commit — and the permission prompt is back. Nothing broke. allowed-tools did exactly what it is specified to do. It is just narrower than the name suggests. The one-line version allowed-tools pre-approves the listed tools…

dev.to ·11h ago

Claude Code's allowed-tools feature is a one-turn grant, not a sandbox. When you add allowed-tools, such as Bash(git add *) or Bash(git commit *), to a skill and invoke it using /commit, Claude stages and commits without asking permission. However, Claude's pre-approval of specific tools expires when you send your next message. The skill remains active, but Claude loses the pre-approval, causing it to prompt for permission again.

The allowed-tools field only specifies which tools do not require a prompt during the current turn, not which tools are restricted. Tools not listed still follow the user's normal permission rules. To remove a tool's permission across all skills, use disallowed-tools. Claude's allowed-tools feature can be applied to bundled scripts within a skill, ensuring the rules match the exact command used by the script.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.