You'll noticeTier: how much it should matter to you
3Useful: my rating, 1 to 5
3Signal: worth watching, 1 to 5
SkillsArea: what it touches
ImprovementsKind: in v2.1.304,
ImprovementsSection of the release
What
When the /claude-api skill loads, it now sets up a PreToolUse hook, a check that runs before a tool call and can block it. The check refuses:
ant commands carrying an ID that contains a staging_, local_, vrtx_ or bdrk_ part.
managed-agents-secure commands that were changed from the ones given in the guide.
When a command is refused, Claude Code tells you it did not run it and why. The skill also does not allow one tool, and if the check cannot run, the skill is not available at all. Claude Code records each refusal.
Why
This limits what the skill's commands can do, so commands that point at the wrong place or stray from the guide are stopped before they run.
The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.
Confirmed sinceAnthropic's documentation has since written up /claude-api, on Feature availability.Desktop is the partial exception: [gateway routing can be configured in the app or by an administrator](/docs/en/llm-gateway-connect#desktop-app). [Claude Desktop on 3P](https://claude.com/docs/third-party/claude-desktop/overview) also run…feature-availabilitysee the edit
How sure we are
Two sources agreeTwo things we can check say the same as this entry.
Anthropic's documentation agreesAnthropic's documentation has since written up /claude-api, on Feature availability.
Anthropic's release notes agreeAdded /claude-api managed-agents-secure, which compares a Managed Agents agent's settings with Anthropic's published security guide, without…
