Skip to main content
Context & actions shows the inventory your agent declared and the runtime evidence associated with its actions. It answers what the agent is configured to do; Traces show what it reported doing.

​
Inspect the inventory

Mission is available only for supported, recognized capability profiles. Other agents open their Actions view. MCP servers show the tools relevant to their narrower connection contract.

​
Choose a useful action to control

Start from actions associated with recorded human intervention or a meaningful risk. Inspect the tool schema and related traces, then propose or write a Rule for a specific decision. Risk classification is guidance: it never creates a Rule or enables enforcement automatically. Use Traces to inspect the actual case behind a counter. A recorded human review, a matching Rule and a blocked call are different events and should not be treated as interchangeable measures.

​
Keep the declaration current

Sync the inventory when the agent starts, or use the repository’s existing rippletide:sync script through rippletide connect. Re-run after changing tools or trusted context. The inventory reflects the last successful declaration; it cannot discover undeclared code automatically. A changed contract disables active Rule releases that depend on it. Unrelated catalogue changes do not disable every release. Review and recompile affected Rules before reactivating them.

​
Configuration versus proof

Visible inventory proves declaration, not a successful live run. SDK agents also need heartbeat, expected tool usage, turn and response-delivery evidence. MCP and Codex have different acceptance contracts. Follow your connection guide and inspect rippletide events --wait before calling the runtime verified.