always_allow, always_ask, and Now auto: Claude's Managed Agents Grows a Middle Option
Anthropic's claude-api skill documents a new auto permission policy for Managed Agents that runs low-risk tool calls, denies high-risk ones automatically, and pauses for human approval only when the risk is indeterminate.
What it is
An update to Anthropic's claude-api Claude Agent Skill documenting a third Managed Agents permission policy alongside always_allow and always_ask.
What it does
The `auto` policy evaluates each tool call and produces one of three outcomes: it runs, it's denied as high-risk with an error tool result while the session keeps going, or it pauses for approval when the evaluation is indeterminate. The commit documents the new `evaluated_permission` and `evaluation` fields on tool-use events and explicitly warns this is not a substitute for a human checkpoint. See the commit for the config example.
Why it matters
Unattended Managed Agents deployments previously had to choose between blanket auto-approval or halting on every ambiguous call; a graded middle policy lets low-risk work continue while genuinely uncertain actions still get a human in the loop.
Netresearch Agentic Skills Lands in awesome-claude-code
Netresearch Agentic Skills is an open-source skills package now listed in the awesome-claude-code directory under Open Source Software.
Session Budgets, Geo Pinning, Multiagent Caps: Managed Agents Grows Up
Anthropic's claude-api skill documents an August Managed Agents update adding session and deployment spend budgets, inference geo pinning, and a multiagent shared usage cap.
Effort Levels, Session Versioning, and initial_events Land in Claude's API Skill
Anthropic updated the claude-api Claude Agent Skill to document five new Managed Agents features, including a model effort level, optimistic-concurrency versioning, and a combined initial_events call.