RFC 788: Bedrock AgentCore Memory L2 construct #825
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
dineshSajwan
changed the title
Abogical
reviewed
Oct 15, 2025
|
|
||
| | Name | Type | Required | Description | | ||
| |------|------|----------|-------------| | ||
| | `name` | `string` | No | The name of the memory | |
Member
There was a problem hiding this comment.
Name is required on the Cloudformation Name property. If the name will be autogenerated on the L2, can you describe how the name will be generated?
| | Name | Type | Required | Description | | ||
| |------|------|----------|-------------| | ||
| | `name` | `string` | No | The name of the memory | | ||
| | `expirationDays` | `Duration` | No | Short-term memory expiration in days (between 7 and 365) | |
Member
There was a problem hiding this comment.
Is there a default for the expirationDays if not specified?
Comment on lines
132
to
283
| ## Memory with Built-in Strategies | ||
|
|
||
| The library provides three built-in LTM strategies: | ||
|
|
||
| 1. **Summarization Strategy** (`MemoryStrategy.BUILT_IN_SUMMARIZATION`) | ||
| - Extracts concise summaries to preserve critical context and key insights | ||
| - Namespace: `/strategies/{memoryStrategyId}/actors/{actorId}/sessions/{sessionId}` | ||
|
|
||
| 2. **Semantic Memory Strategy** (`MemoryStrategy.BUILT_IN_SEMANTIC`) | ||
| - Extracts general factual knowledge, concepts and meanings from raw conversations | ||
| - Namespace: `/strategies/{memoryStrategyId}/actors/{actorId}` | ||
|
|
||
| 3. **User Preference Strategy** (`MemoryStrategy.BUILT_IN_USER_PREFERENCE`) | ||
| - Extracts user behavior patterns from raw conversations | ||
| - Namespace: `/strategies/{memoryStrategyId}/actors/{actorId}` | ||
|
|
||
| ```typescript | ||
| import * as cdk from "aws-cdk-lib"; | ||
| import { | ||
| Memory, | ||
| MemoryStrategy, | ||
| } from "aws-cdk/bedrock-agentcore-alpha/memory"; | ||
|
|
||
| // Create memory with built-in strategies | ||
| const memory = new Memory(this, "MyMemory", { | ||
| name: "my_memory", | ||
| description: "Memory with built-in strategies", | ||
| expirationDays: cdk.Duration.days(90), | ||
| memoryStrategies: [ | ||
| MemoryStrategy.BUILT_IN_SUMMARIZATION, | ||
| MemoryStrategy.BUILT_IN_SEMANTIC, | ||
| MemoryStrategy.BUILT_IN_USER_PREFERENCE, | ||
| ], | ||
| }); | ||
| ``` | ||
|
|
||
| ## Memory with Built-in Strategies - Custom Namespace | ||
|
|
||
| You can customise the namespace, i.e. where the memories are stored by using the following methods: | ||
|
|
||
| 1. **Summarization Strategy** (`MemoryStrategy.fromBuiltInSummarization(props)`) | ||
| 2. **Semantic Memory Strategy** (`MemoryStrategy.fromBuiltInSemantic(props)`) | ||
| 3. **User Preference Strategy** (`MemoryStrategy.fromBuiltInUserPreference(props)`) | ||
|
|
||
| ```typescript | ||
| import * as cdk from "aws-cdk-lib"; | ||
| import { | ||
| Memory, | ||
| MemoryStrategy, | ||
| } from "aws-cdk/bedrock-agentcore-alpha/memory"; | ||
|
|
||
| // Create memory with built-in strategies using custom namespaces | ||
| const memory = new Memory(this, "MyMemory", { | ||
| name: "my_memory", | ||
| description: "Memory with built-in strategies", | ||
| expirationDays: cdk.Duration.days(90), | ||
| memoryStrategies: [ | ||
| MemoryStrategy.fromBuiltInUserPreference({ | ||
| name: "CustomerPreferences", | ||
| namespaces: ["support/customer/{actorId}/preferences"] | ||
| }), | ||
| MemoryStrategy.fromBuiltInSemantic({ | ||
| name: "CustomerSupportSemantic", | ||
| namespaces: ["support/customer/{actorId}/semantic"] | ||
| }), | ||
| MemoryStrategy.fromBuiltInSummarization({ | ||
| name: "SessionSummaries", | ||
| namespaces: ["support/sessions/{sessionId}/summary"] | ||
| }), | ||
| ], | ||
| }); | ||
| ``` | ||
|
|
||
| ## Custom Strategies | ||
|
|
||
| You can also create custom memory strategies using your specified models and prompts. | ||
| According to the strategy that you will be customizing, you will have to specify extraction and/or consolidation FMs | ||
| and prompt templates to append to the system prompt of the memory strategy. You can do so by using: | ||
|
|
||
| 1. **Summarization Strategy** (`MemoryStrategy.fromCustomSummaryOverride(props)`) | ||
| 2. **Semantic Memory Strategy** (`MemoryStrategy.fromCustomSemanticOverride(props)`) | ||
| 3. **User Preference Strategy** (`MemoryStrategy.fromCustomUserPreferenceOverride(props)`) | ||
|
|
||
| Since a custom strategy requires you to invoke certain FMs, you need a role with appropriate permissions. For that, you can: | ||
|
|
||
| - Let the L2 construct create a minimum permission role for you when use L2 Bedrock Foundation Models. | ||
| - Use a custom role with the overly permissive `AmazonBedrockAgentCoreMemoryBedrockModelInferenceExecutionRolePolicy` managed policy. | ||
| - Use a custom role with your own custom policies. | ||
|
|
||
| ### Memory with Custom Execution Role | ||
|
|
||
| Keep in mind that memories that **do not** use custom strategies do not require a service role. So even if you provide it, | ||
| it will be ignored as it will never be used. | ||
|
|
||
| ```typescript | ||
| import * as cdk from "aws-cdk-lib"; | ||
| import * as iam from "aws-cdk-lib/aws-iam"; | ||
| import { Memory } from "aws-cdk/bedrock-agentcore-alpha/memory"; | ||
|
|
||
| // Create a custom execution role | ||
| const executionRole = new iam.Role(this, "MemoryExecutionRole", { | ||
| assumedBy: new iam.ServicePrincipal("bedrock-agentcore.amazonaws.com"), | ||
| managedPolicies: [ | ||
| iam.ManagedPolicy.fromAwsManagedPolicyName( | ||
| "AmazonBedrockAgentCoreMemoryBedrockModelInferenceExecutionRolePolicy" | ||
| ), | ||
| ], | ||
| }); | ||
|
|
||
| // Create memory with custom execution role | ||
| const memory = new Memory(this, "MyMemory", { | ||
| name: "my_memory", | ||
| description: "Memory with custom execution role", | ||
| expirationDays: cdk.Duration.days(90), | ||
| executionRole: executionRole, | ||
| }); | ||
| ``` | ||
|
|
||
| ### Custom Strategy Example | ||
|
|
||
| ```typescript | ||
| import * as cdk from "aws-cdk-lib"; | ||
| import * as bedrock from "@aws-cdk/aws-bedrock-alpha"; | ||
| import { | ||
| Memory, | ||
| MemoryStrategy, | ||
| MemoryStrategyType, | ||
| } from "aws-cdk/bedrock-agentcore-alpha/memory"; | ||
|
|
||
| // Create a custom semantic memory strategy | ||
| const customSemanticStrategy = MemoryStrategy.fromCustomSemanticOverride({ | ||
| name: "custom-semantic-strategy", | ||
| description: "Custom semantic memory strategy", | ||
| namespaces: ["/custom/strategies/{memoryStrategyId}/actors/{actorId}"], | ||
| customConsolidation: { | ||
| model: "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0", | ||
| customPrompt: "Custom consolidation prompt for semantic memory", | ||
| }, | ||
| customExtraction: { | ||
| model: "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0", | ||
| customPrompt: "Custom extraction prompt for semantic memory", | ||
| }, | ||
| }); | ||
|
|
||
| // Create memory with custom strategy | ||
| const memory = new Memory(this, "MyMemory", { | ||
| name: "my-custom-memory", | ||
| description: "Memory with custom strategy", | ||
| expirationDays: cdk.Duration.days(90), | ||
| memoryStrategies: [customSemanticStrategy], | ||
| }); | ||
| ``` |
Member
There was a problem hiding this comment.
I think this interface could be simplified with one function with default props.
Without custom namespaces:
const memory = new Memory(this, "MyMemory", {
name: "my_memory",
description: "Memory with built-in strategies",
expirationDays: cdk.Duration.days(90),
memoryStrategies: [
MemoryStrategy.userPreference(),
MemoryStrategy.semantic(),
MemoryStrategy.summarization(),
],
});
With custom strategies:
const memory = new Memory(this, "MyMemory", {
name: "my_memory",
description: "Memory with built-in strategies",
expirationDays: cdk.Duration.days(90),
memoryStrategies: [
MemoryStrategy.userPreference({
name: "CustomerPreferences",
namespaces: ["support/customer/{actorId}/preferences"]
}),
MemoryStrategy.semantic({
name: "CustomerSupportSemantic",
namespaces: ["support/customer/{actorId}/semantic"]
}),
MemoryStrategy.summarization({
name: "SessionSummaries",
namespaces: ["support/sessions/{sessionId}/summary"]
}),
],
});
Custom memory strategy:
const customSemanticStrategy = MemoryStrategy.semantic({
name: "custom-semantic-strategy",
description: "Custom semantic memory strategy",
namespaces: ["/custom/strategies/{memoryStrategyId}/actors/{actorId}"],
customConsolidation: {
model: "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0",
customPrompt: "Custom consolidation prompt for semantic memory",
},
customExtraction: {
model: "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0",
customPrompt: "Custom extraction prompt for semantic memory",
},
});
Abogical
requested changes
Oct 15, 2025
1 task
Abogical
requested changes
Oct 17, 2025
| memoryName: "my_memory", | ||
| description: "Memory with custom execution role", | ||
| expirationDays: cdk.Duration.days(90), | ||
| // executionRole: executionRole, |
Member
There was a problem hiding this comment.
This line shouldn't be a comment.
Abogical
requested changes
Oct 17, 2025
Abogical
requested changes
Oct 17, 2025
|
|
||
| ### Are there any open issues that need to be addressed later? | ||
|
|
||
| Self managed strategy is not available yet, it will be added as a long term memory strategy. |
Member
There was a problem hiding this comment.
Suggested change
| Self managed strategy is not available yet, it will be added as a long term memory strategy. | |
| The [Self-managed memory strategy](https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/memory-self-managed-strategies.html) does not have an exposed CloudFormation interface yet. It will be added later as a long term memory strategy once it is exposed. |
Abogical
requested changes
Oct 17, 2025
| - For Semantic:`semantic_builtin_<suffix>` | ||
| - For User Preferences: `preference_builtin_<suffix>` | ||
|
|
||
| Where the suffix is a 5 characters string ([a-z, A-Z, 0-9]). |
Member
There was a problem hiding this comment.
suffix will be generated via the Names.uniqueId(scope) using the memory's scope as discussed in slack.
Abogical
previously requested changes
Oct 17, 2025
alvazjor
dismissed
Abogical’s stale review
Discussed and agreed on the proposal
alvazjor
added
pr/do-not-merge
Contributor
|
Signing off the RFC. Will keep it open for the final comments period. If no additional comments are added, this can be merged on October 28th |
alvazjor
approved these changes
Oct 21, 2025
Member
|
No comments made. Merging. |
mergify bot
pushed a commit
to aws/aws-cdk
that referenced
this pull request
Oct 29, 2025
### Issue # (if applicable) Related to aws/aws-cdk-rfcs#825 ### Reason for this change Adding a new alpha package for Amazon Bedrock AgentCore and add support for memory. ### Description of changes - Create a new alpha package - Add L2 constructs for memory - Add documentation - Add tests ### Describe any new or updated permissions being added Using permissions for agent core defined in https://docs.aws.amazon.com/service-authorization/latest/reference/list_amazonbedrockagentcore.html ### Description of how you validated changes Unit tests, integration tests, manual tests ### Checklist - [x] My code adheres to the [CONTRIBUTING GUIDE](https://github.com/aws/aws-cdk/blob/main/CONTRIBUTING.md) and [DESIGN GUIDELINES](https://github.com/aws/aws-cdk/blob/main/docs/DESIGN_GUIDELINES.md) ---- *By submitting this pull request, I confirm that my contribution is made under the terms of the Apache-2.0 license*
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is a request for Amazon Bedrock AgentCore L2 construct that simplifies the deployment and management of AI agents at scale by wrapping the AgentCore L1 constructs. It provides a high-level, object-oriented approach to creating and managing Amazon Bedrock AgentCore resources. This enables developers to deploy highly effective agents securely using any framework and model. The package will provide constructs for the following primitives:
AgentCore Memory
See # #785 for
additional details.
APIs are signed off by @{BAR_RAISER}.
By submitting this pull request, I confirm that my contribution is made under
the terms of the Apache-2.0 license