developing-agentforce

To build and deploy Agentforce agents, you author Agent Script bundles (.agent) and AiAuthoringBundle metadata, wire backing logic via Apex, Flows, or Prompt Templates, then validate, preview, and publish using strict sf agent CLI commands.

Agent Script compilation issues often stem from incorrect AiAuthoringBundle syntax, missing backing logic wiring, or invalid sf agent CLI flags. You must validate metadata changes, ensure Agent User prerequisites are met, and use trace-grounded debugging to diagnose errors.

Debugging Agentforce behavior requires analyzing live preview traces to identify execution failures. The approach enforces trace-grounded analysis, checking validation outputs and environment prerequisites before modifying Agent Specs or backing logic.

Yes, you can wire Apex classes, Flows, and Prompt Templates as backing logic for Agentforce actions. The skill guides integration, handles compilation prerequisites, and ensures environment validation passes before deploying the AiAuthoringBundle.

Before creating an Agentforce agent, you need the sf CLI configured with --json output, an environment with Agent User prerequisites satisfied, and an approved Agent Spec. Validation and environment checks are enforced prior to any code generation or deployment.

Agent Script authoring bundles (.agent) are ideal when you need version-controlled, metadata-driven Agentforce development with end-to-end validation, trace-based debugging, and structured deployment via sf agent CLI commands rather than manual UI configuration.