AWS bedrock-agentcore documentation change

Service: bedrock-agentcore · 2026-06-19 · Documentation high

File: bedrock-agentcore/latest/devguide/recommendations-system-prompt.md

Summary

Expanded trace sources, added KMS encryption option, and added instructions for asynchronous job handling with --wait flag and result retrieval

Security assessment

Added --kms-key option for encrypting recommendation results using customer-managed keys, enhancing data protection. No evidence of addressing a specific security vulnerability.

Diff

diff --git a/bedrock-agentcore/latest/devguide/recommendations-system-prompt.md b/bedrock-agentcore/latest/devguide/recommendations-system-prompt.md
index 13c92a2a7..1cf18dbdc 100644
--- a//bedrock-agentcore/latest/devguide/recommendations-system-prompt.md
+++ b//bedrock-agentcore/latest/devguide/recommendations-system-prompt.md
@@ -24 +24 @@ AgentCore CLI
-The CLI accepts two trace sources and three system prompt input modes. Combine them as needed:
+The CLI accepts several trace sources and three system prompt input modes. Combine them as needed:
@@ -26 +26 @@ The CLI accepts two trace sources and three system prompt input modes. Combine t
-  * **Trace sources:** CloudWatch Logs (`--lookback`) or inline spans (`--spans-file`)
+  * **Trace sources:** CloudWatch Logs (`--lookback`), inline spans (`--spans-file`), a local insights run (`--from-insights <id>` — uses a local insights run as trace source; resolves the batch eval ARN), or a batch evaluation ARN directly (`--batch-evaluation-arn <arn>` — uses a batch evaluation ARN directly as trace source)
@@ -31,0 +32,2 @@ The CLI accepts two trace sources and three system prompt input modes. Combine t
+  * **Optional encryption:** a KMS key (`--kms-key <arn>` — KMS key ARN for encrypting recommendation results)
+
@@ -40 +42,4 @@ The CLI accepts two trace sources and three system prompt input modes. Combine t
-      --lookback 7
+      --lookback 7 \
+      --wait
+
+`run recommendation` starts an asynchronous job and returns only a `recommendationId` and an initial `PENDING` or `IN_PROGRESS` status. Add `--wait` to block until the recommendation reaches a terminal state. To retrieve the completed result later, see Retrieve the result.
@@ -87,0 +93,6 @@ The CLI auto-resolves the full JSON path from the agent runtime ARN’s `configu
+**Retrieve the result:**
+
+Use `view recommendation` with the recommendation job ID to fetch the completed result. Add `--json` for machine-readable output containing `recommendedSystemPrompt` and `explanation`.
+    
+        agentcore view recommendation <recommendation-id> --json
+