awslabs/agentcore-samples

Summary I was trying https://github.com/awslabs/agentcore-samples/tree/main/01-features/01-harness/01-advanced-examples/05-agent-skills but in oetl_logs I could see agentcore harness was unable to load skills/xlsx Details Followed the steps in the tutorial. " "command": "ls -la .agents/skills/xlsx/ 2>/dev/null && echo 'skill directory OK' || echo 'skill not found'"" command gave me skill directory ok but in logs I could see it failed to load it. 2026-05-27T19:17:28.123Z2026-05-27 19:17:27,974 WARNING [strands.vended_plugins.skills.agent_skills] [agent_skills.py:334] [trace_id=xxx span_id=xx resource.service.name=harness_SkillsDemo_xxxx.DEFAULT trace_sampled=True] - path=</home/.agents/skills/xlsx> \| skill source path does not exist, skipping   | 2026-05-27T19:17:28.124Z2026-05-27 19:17:28,124 WARNING [strands.vended_plugins.skills.agent_skills] [agent_skills.py:110] [trace_id=xxxx span_id=xxx resource.service.name=harness_SkillsDemo_xxx.DEFAULT trace_sampled=True] - no skills were l

Summary Briefly describe the problem or the help you need. When creating an AgentCore registry with authorizerType: CUSTOM_JWT, the Search Records tab does not appear in the AWS console. The same registry created with AWS_IAM shows both Managed Records and Search Records tabs. Details Provide more details about your issue, including steps you’ve tried and what didn’t work. - Created registry via boto3 SDK using authorizerType: "CUSTOM_JWT" with valid authorizerConfiguration (OIDC discovery URL + custom claims) - The same JWT discovery URL and tenant ID work correctly for runtime and gateway auth - Registry status shows as ACTIVE - Console only displays the "Managed Records" tab - When creating with authorizerType: "AWS_IAM", both "Managed Records" and "Search Records" tabs appear immediately Expected Behavior What outcome were you hoping for? Search Records tab should be visible in the console for registries created with CUSTOM_JWT authorizer, same as with AWS_IAM. Add

Summary Having trouble understanding how to deploy Microsoft Agent Framework in .NET (though it could be Python for the sake of the tutorial) onto AWS Bedrock AgentCore. Our work environment is we are a Microsoft shop for tooling and language but build on AWS. Just looking for a very simple example of a Microsoft Agent Framework (MAF) agent in .NET or Python being deployed and running within AWS Bedrock AgentCore Details Would love even a minimal working sample I could work through, understand and build on top of.

Summary Briefly describe the problem or the help you need. _Example: Trouble setting up local dev environment_ Details Provide more details about your issue, including steps you’ve tried and what didn’t work. _Example: I ran `npm install` but the build fails with error XYZ_ Expected Behavior What outcome were you hoping for? _Example: Expected server to start after running `npm start`_ Additional Context Add any other information, screenshots, logs, or references that could help.

Summary When I am trying to launch the agentcore_runtime within sagemaker studio "launch_result = agentcore_runtime.launch()", I get the below AccessDeniedException error message. Details IAM policy for SageMakerExecution Role: { "Version": "2012-10-17", "Statement": [ { "Sid": "ExistingS3Access", "Effect": "Allow", "Action": [ "s3:ListBucket", "s3:GetObject", "s3:PutObject", "s3:DeleteObject", "s3:PutLifecycleConfiguration", "s3:GetLifecycleConfiguration" ], "Resource": [ "arn:aws:s3:::SageMaker", "arn:aws:s3:::SageMaker/", "arn:aws:s3:::bedrock-agentcore-codebuild-sources-" ] }, { "Sid": "BedrockAgentCoreRuntime", "Effect": "Allow", "Action": [ "bedrock-agentcore:CreateAgentRuntime", "bedrock-agentcore:GetAgentRuntime", "bedrock-agentcore:DeleteAgentRuntime", "bedrock-agentcore:UpdateAgentRuntime" ], "Resource": "" }, { "Sid": "SageMakerIAMSetup", "Effect": "Allow", "Action": [ "iam:CreateRole", "iam:PutRolePolicy", "iam:CreatePolicy", "iam:AttachRolePolicy", "iam:GetRole", "iam:Get