The sf-ai-agentscript skill enables developers to build, debug, and deploy Salesforce Agentforce agents using a code-first approach via the Agent Script DSL. By moving from prompt-based suggestions to code-enforced guarantees, this skill allows for the creation of robust agents within a single .agent file. It provides specialized guidance on Finite State Machine (FSM) architecture, syntax validation, and instruction resolution, while helping developers navigate critical constraints such as API v65.0 requirements, reserved field names, and the avoidance of nested conditionals. It is the definitive resource for Salesforce developers aiming for high-reliability, production-grade agentic AI.
Características Principales
01Automated .agent DSL syntax validation and guardrail checks
02Deterministic FSM (Finite State Machine) architecture for Agentforce
03Pre-deployment checks for API v65.0+ and Einstein Agent User validation
04Mapping tables for complex data types and SObject record types
0537 GitHub stars
06Support for hybrid reasoning and complex instruction resolution