SimuLearn AgentCore Memory DIY validator fails despite records existing - "No long-term memory records found for DIY customer"

Question

I've completed the AgentCore Memory lab DIY challenge three times and the validator consistently fails with: "No long-term memory records (preferences, facts, or summaries) were found for the DIY customer."
I've confirmed records exist via retrieve-memory-records — 10 preferences, 10 facts, and multiple summaries for customer-456 in both the console and CloudFormation memory resources. The Lambda returns correct memory context (long_term_preferences: 10, long_term_facts: 10) via API Gateway. Streamlit shows correct personalized responses for both customers.
The issue appears to be that records are stored under /strategies/{strategyId}/actors/{actorId}/ namespaces (as AgentCore creates them), but a broad search with namespace prefix / returns empty results. The validator appears to use a broad namespace query that doesn't traverse into /strategies/ paths.
Memory IDs: agentcore_memory_cfn_comparison-jnNKhA3tBS, agentcore_memory_customer_service-hwpPdhD6sO
Is this a known validator bug? Is there a workaround to get the validator to pass, or can completion credit be granted manually?

Florian Turnwald · Accepted Answer

Sounds like this is a classic lab environment issue, not a problem with your implementation. Your analysis regarding the namespace mismatch sounds correct.

Lab validators often rely on rigid, hardcoded query paths. The validator script was likely written for an earlier version or simpler schema. Because AgentCore stores the records in a nested namespace (/strategies/{strategyId}/actors/{actorId}/), and the validator is performing a non-recursive broad search at the root (/), it fails to "see" your perfectly valid data.
Since you cannot modify the lab's underlying validation script, here is how you can proceed:

### 1. Manual Namespace Override (Workaround)

If the lab environment provides you with sufficient IAM permissions and CLI/API access, you can attempt to manually inject a dummy set of memory records directly into the root / namespace. This often tricks the rigid validator into passing your environment.

### 2. Request Manual Completion Credit

If direct API manipulation to the root namespace isn't possible or fails, this is officially a bug in the lab's validation logic. You will need to reach out to the lab platform's support team (e.g., via the AWS Skill Builder support widget or event platform). Provide them with the excellent debugging details you posted here, along with your Memory IDs (agentcore_memory_cfn_comparison-jnNKhA3tBS, agentcore_memory_customer_service-hwpPdhD6sO), to request a manual override and completion credit.

SimuLearn AgentCore Memory DIY validator fails despite records existing - "No long-term memory records found for DIY customer"

1. Manual Namespace Override (Workaround)

2. Request Manual Completion Credit

新增您的答案

相關內容