jaganpro/sf-skills
Overview
This skill describes the standard Salesforce Agentforce platform workflow for building agents using the Setup UI (Agentforce Builder). It covers topic and action configuration, GenAiFunction/GenAiPlugin metadata, PromptTemplate authoring, Einstein Models API usage from Apex, and custom Lightning types for rich inputs/outputs. Use this when you prefer declarative, point-and-click agent development rather than code-first Agent Script DSL.
How this skill works
Configure agents in Setup → Agentforce → Agents by creating topics, writing instructions, and adding actions that map to Flows, Apex InvocableMethods, or PromptTemplates. Register each executable as a GenAiFunction, optionally group functions into a GenAiPlugin, then publish the agent so the Agentforce runtime can plan and invoke actions. For custom AI logic, call Einstein Models API from Apex; for rich UI inputs/outputs, create LightningTypeBundle bundles.
When to use it
- Building agents via the Salesforce Setup UI without authoring .agent files
- Mapping Autolaunched Flows or @InvocableMethod Apex to agent actions
- Reusing prompts across agents and Flows with PromptTemplate metadata
- Creating rich input forms or formatted outputs with custom Lightning types
- Registering grouped actions using GenAiPlugin for organization
Best practices
- Deploy Flow/Apex/Prompt targets first, then deploy GenAiFunction, then GenAiPlugin, then publish the agent
- Write explicit, scoped topic descriptions and instructions to guide the LLM planner
- Define required inputs and mark displayable outputs so the agent collects slots cleanly
- Use PromptTemplate variable names exactly (case-sensitive) when binding fields
- Use Einstein Models API in async jobs (Queueable/Batch) for heavy or bulk tasks
Example use cases
- Service agent that looks up order status: Flow target, GenAiFunction wrapper, topic routed by planner
- Employee agent that runs a payroll validation Apex @InvocableMethod via GenAiFunction
- Knowledge agent that generates personalized replies using a PromptTemplate invoked by an action
- Order management plugin grouping functions: lookup, cancel, and return as a GenAiPlugin
- Custom product detail input form using LightningTypeBundle editor and renderer
FAQ
Always deploy the underlying Flow/Apex/PromptTemplate first. GenAiFunction references must point to active targets.
When should I use Einstein Models API instead of a PromptTemplate?
Use the Models API for custom, programmatic LLM calls from Apex (async patterns, batching, or advanced orchestration). Use PromptTemplates for reusable authoring inside the platform and Flows.
16 skills
This skill analyzes Salesforce debug logs to detect governor limits and performance bottlenecks, then suggests actionable fixes.
This skill helps design Salesforce Agentforce conversation flows with persona, topic architectures, utterance libraries, and escalation guardrails for
This skill extracts and analyzes Agentforce session tracing data from Salesforce Data 360 at scale, enabling debugging with Polars and Parquet storage.
This skill executes dual-track agent testing with multi-turn API tests and CLI validation, analyzes coverage, and automatically fixes issues for faster quality.
This skill generates Salesforce ERD diagrams and LWC mockups with Nano Banana Pro, plus parallel Gemini code reviews and research for faster development.
This skill helps you generate and review Salesforce Apex code following 2025 best practices and a 150-point scoring system.
This skill helps you design and validate Salesforce flows with Winter 26 best practices, ensuring reliable, bulk-safe deployments.
This skill helps you manage Salesforce data efficiently by generating and executing SOQL, DML, and bulk operations with test data and cleanup commands.
This skill automates Salesforce DevOps deployments using sf CLI v2, guiding object-first deployments, flow activation, and scratch org management for reliable
This skill helps you design and implement secure Salesforce integrations with Named Credentials, External Services, and event-driven patterns.
This skill greets you, explains its structure, and educates on Claude Code components through a minimal hello-world example.
This skill guides you to write, debug, test, and deploy Agent Script DSL agents with deterministic FSM architecture for Salesforce Agentforce.
This skill helps you generate and validate Salesforce Connected Apps and External Client Apps with secure OAuth configuration and scoring.
This skill generates production-ready Salesforce architecture diagrams in Mermaid with ASCII fallback, auto-discovering metadata and validating diagram quality.
This skill helps you configure and manage UI-driven Agentforce agents with topics, actions, and prompts for declarative development.
This skill helps maintain legacy Agentforce agents by generating and validating Agent Script based configurations, topics, and actions with structured scoring.
