aws / aws/bedrock-agentcore-starter-toolkit
AgentCore Deployment Failure: ResourceNotFoundException Due to Stale Agent Configuration
- Dominant language
- Python
- Stars
- 508
- Forks
- 155
- Avg merge
- 8h 50m
- Merged PRs (30d)
- 4
Description
## Summary
Agent deployment fails with `ResourceNotFoundException` when trying to update a non-existent agent ID that remains in the local configuration file after the agent was deleted from AgentCore.
## Environment
- **AgentCore SDK Version**: bedrock-agentcore-starter-toolkit==0.1.25
- **Platform**: macOS
- **Region**: us-east-1
- **Framework**: Strands Agents
- **Python**: 3.13
- **Deployment Method**: CodeBuild (cloud build)
- **Container Runtime**: ARM64
## Issue Description
When deploying an agent, the process fails during the "Deploying to Bedrock AgentCore..." step with:
```
launch_result = agentcore_runtime.launch()
✅ ECR repository available: .dkr.ecr.us-east-1.amazonaws.com/bedrock-agentcore-strands_claude_getting_started
Using execution role from config: arn:aws:iam:::role/role-agentcore
Preparing CodeBuild project and uploading source...
Using CodeBuild role from config: arn:aws:iam:::role/role-codebuild
Using .dockerignore with 44 patterns
Uploaded source to S3: strands_claude_getting_started/source.zip
Updated CodeBuild project: bedrock-agentcore-strands_claude_getting_started-builder
Starting CodeBuild build (this may take several minutes)...
Starting CodeBuild monitoring...
🔄 QUEUED started (total: 0s)
✅ QUEUED completed in 1.3s
🔄 PROVISIONING started (total: 2s)
✅ PROVISIONING completed in 9.3s
🔄 DOWNLOAD_SOURCE started (total: 11s)
✅ DOWNLOAD_SOURCE completed in 1.3s
🔄 BUILD started (total: 12s)
✅ BUILD completed in 17.2s
🔄 POST_BUILD started (total: 29s)
✅ POST_BUILD completed in 11.8s
🔄 COMPLETED started (total: 41s)
✅ COMPLETED completed in 1.3s
🎉 CodeBuild completed successfully in 0m 42s
CodeBuild completed successfully
✅ CodeBuild project configuration saved
Deploying to Bedrock AgentCore...
Failed to update agent ID 'strands_claude_getting_started-af637': An error occurred (ResourceNotFoundException) when calling the UpdateAgentRuntime operation: Agent 'strands_claude_getting_started-af637' was not found. Please check the agent ID and try again.
```
## Root Cause
The `.bedrock_agentcore.yaml` configuration file contains a stale `bedrock_agentcore` section with an agent ID that no longer exists in the AgentCore service. The deployment process attempts to **update** the existing agent instead of creating a new one.
## Steps to Reproduce
1. Deploy an agent successfully (creates agent ID in config)
2. Delete the agent from AgentCore (via console/API) OR agent gets deleted/corrupted
3. Attempt to redeploy the same agent using `agentcore_runtime.launch()`
4. CodeBuild completes successfully (container builds and pushes to ECR)
5. Deployment fails during "Deploying to Bedrock AgentCore..." step with "Agent not found" error
## Expected Behavior
- The SDK should detect that the agent ID no longer exists
- Automatically create a new agent instead of trying to update
- Update the configuration with the new agent ID
Contributor guide
Research direction
Start at the agentcore_runtime.launch() deployment flow and inspect how the bedrock_agentcore section in .bedrock_agentcore.yaml selects update versus create. Trace the UpdateAgentRuntime failure for a missing agent ID, then verify that a redeploy creates a replacement agent and updates the configuration after the CodeBuild step succeeds.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- aws, python
- Domain
- cloud, devops
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Clearly specified
- Newbie friendliness
- 45/100