Skip to content
Notifications
Clear all

How do I stop Q from suggesting AWS proprietary libraries all the time?

1 Posts
1 Users
0 Reactions
0 Views
(@devops_not_grunt)
Reputable Member
Joined: 5 months ago
Posts: 262
Topic starter   [#23911]

Let's be clear: Amazon Q isn't an AI coding assistant, it's an AWS land-grab tool with a code completion feature. I asked it to generate a simple function to put an item into a DynamoDB table, and it defaulted to using the **AWS SDK for Java 2.x**, which is fine, but then it *aggressively* suggested I refactor my entire service layer to use their proprietary **AWS Enhanced DynamoDB Client** and **DynamoDB Mapper** annotations.

Here's the pattern I'm talking about. You ask for something standard:

```java
// Me: "Generate a method to save a user profile to DynamoDB"
public void saveUserProfile(UserProfile profile) {
PutItemRequest putItemRequest = PutItemRequest.builder()
.tableName("UserProfiles")
.item(ItemMapper.toAttributeMap(profile))
.build();
dynamoDbClient.putItem(putItemRequest);
}
```

And then the suggestions start. "Consider using the Enhanced Client for type-safe operations..." followed by a multi-line rewrite that locks you into their specific annotation model. It does this for S3 (use the Transfer Manager), SQS (use the `QueueMessagingTemplate` from Spring Cloud AWS), and of course, any event-driven architecture gets funneled toward EventBridge and Step Functions.

The problem isn't that these tools are badβ€”some are quite good. The problem is the utter lack of architectural neutrality. It's like asking for a screwdriver and having a salesperson hand you one that only works with screws sold by their company, while they quietly suggest you replace all your existing screws.

In our last platform incident, a junior dev followed Q's "optimized" suggestions for a Lambda-to-SQS integration using the proprietary Spring Cloud AWS helpers. When we needed to migrate that workload to a hybrid environment for compliance, the vendor lock-in cost us two weeks of refactoring. Q's "best practice" became our week-long setback.

Has anyone found a prompt incantation or configuration setting that actually tones this down, or is the only solution to treat its suggestions as a constant stream of AWS marketing that you have to consciously filter out?



   
Quote