Agentforce Planner Failing to Identify or Execute SubAgent's Actions

To resolve this issue and prevent future occurrences, please implement the following corrective actions and best practices:

1. Verify Source Metadata and Pipelines: Collaborate with your Release Management team to audit the deployment branches and pipelines for any deprecated, unwanted, or duplicate references to subagent's actions. To establish a reliable baseline, retrieve the metadata directly from the source organization using the Salesforce CLI to ensure no legacy duplicates are present in the local source.

2. Audit the Deployment Package: Prior to execution, thoroughly validate the deployment package to confirm there are no duplicate actions assigned to any Agentforce topics.

3. Validate Parameters Post-Deployment: Deploy the metadata into a clean sandbox environment. Following the deployment, manually verify that all input and output parameters for the respective subagent's actions are accurately populated, as missing parameters can also trigger execution failures.

4. Utilize a Clean Deployment Strategy: If the problem persists, consider establishing a fresh, clean deployment branch. As a diagnostic step, attempt to deploy the configurations using standard, native Salesforce deployment mechanisms (such as Salesforce Extensions for VS Code or traditional Change Sets) to determine if the third-party tool is responsible for injecting the corrupted metadata artifacts.