Skip to content

Examples

Ask your agent about your OpenLatch organization in your own words. Each example below gives a question, the tools your agent calls, and what it answers. You need your agent connected to OpenLatch, with the skills installed so it picks the right tools.

“How many zones do I have?”

Your agent calls zones_read and lists each zone by its path in your org tree, with the agents it holds and the intent policies in force.

“How many agents are in the Payments zone?”

Your agent finds the zone with zones_read, then lists its agents with the same tool, each with its control level and when it was last seen.

“Which intent policies run in Monitor in the Payments zone?”

Your agent finds the zone, then lists its policies with policies_read, filtered by mode. Ask it next which ones are ready for Enforce.

“Write a policy that keeps agents out of production credentials.”

Your agent runs the interview with you and shows you every sentence before it ratifies. Author a policy through your agent walks through it.

“Which of my agents are only Monitored, and which have no owner?”

Your agent lists your agents with agents_read and reads the control level and the owner of each.

“What did my agent do in its last session?”

Your agent finds the session with agent_sessions_read, then reads the execution graph over that session’s time range: the tools it used, the commands it ran and the files it wrote.

“Which agents wrote to config.py today?”

Your agent lists your agents with agents_read, then reads each one’s execution graph for today with agent_sessions_read and keeps the agents with a file action on that path.

“Show me the Security anomalies of the last seven days.”

Your agent lists them with anomalies_read, filtered by dimension and time window, each with its agent, its policy and its verdict.

“Which Asks are waiting for an answer?”

Your agent lists them with asks_read. Tell it how to answer one, and it answers with triage_write.