Help Center
All articlesOperations Command Center

Investigate blocked actions

Review why an action was stopped and decide whether policy, access, or agent behavior should change.

Updated June 19, 20265 min read

A block is a control working

Blocked actions are not noise. They are moments where policy prevented an agent from crossing a boundary. The investigation should decide whether the block was correct, the policy is too broad, or the agent needs a narrower scope. An LLM trace can help when the block involved a model call.

Iron Gorilla operations dashboard with metrics, approval queues, and agent activity.
The operations dashboard gathers active risk, approvals, and audit context in one working view.

Investigating the block

Open the blocked action from Operations or audit history and review the agent, tool, destination, policy, and trust context together. Repeated patterns usually matter more than a single isolated block.

Adjust policy only when the business boundary was wrong or incomplete. Adjust connector or agent scope when the agent had more access than it needed.

Operations Command CenterUse the operations dashboard

Find current risk, waiting approvals, blocked actions, and agent activity in one view.

Policy BuilderSimulate a policy before launch

Use sample actions to see exactly how a policy will behave before it affects live work.

LLM ProxyRead an LLM trace

Understand provider, token, latency, cost, policy, and refusal details for a model call.

Need help?

Tell us where you got stuck.

Share what you were trying to do, what happened, and how urgent it is. We will route it to the right team.

Submit a ticket