Language:JapaneseEnglish

agentforce-adlc Claude Code Plugin

Category
Development
Topics
Enterprise Business Platforms · AI Agents & AI App Development · Security, Auth & Compliance
First cataloged
2026-07-09 (UTC)
Explanation last updated
2026-07-16 (UTC)
Source (GitHub) last updated
2026-07-30 (UTC) (28 days ago)

The explanation below is AI-generated. Please verify it against the sources.

agentforce-adlc is a Claude Code plugin that supports the full development lifecycle for Salesforce Agentforce agents, from initial authoring through production optimization. According to its README, it provides a set of skills that write .agent files directly in the Agent Script DSL, check a Salesforce org for missing Flow/Apex/Retriever targets and scaffold stubs for them, then validate, publish, and activate the agent. It also includes skills for running preview and batch tests (including adversarial safety-probe utterances) and for analyzing production session traces pulled from Data Cloud to find and fix routing or grounding problems. A dedicated skill performs an OWASP LLM Top 10 style security assessment against a live agent. Safety review is described as being embedded at each stage of the lifecycle rather than added afterward.

About the service

Salesforce Agentforce is Salesforce's platform for building and running autonomous AI agents that draw on CRM data and existing business logic (Flows, Apex, APIs). This plugin's skills operate against a connected Salesforce org and its Data Cloud data to author, deploy, test, and refine Agentforce agents.

What you can do with agentforce-adlc

  • Generate .agent files (topics, actions, variables, deterministic logic) from a natural-language description of the desired agent
  • Check a Salesforce org for existing Flow/Apex/Retriever targets and scaffold stub implementations for any that are missing
  • Validate, publish, and activate an authoring bundle to deploy the agent to an org
  • Run preview sessions and batch tests, including automatically generated adversarial safety-probe utterances
  • Pull STDM session traces from Data Cloud to identify routing or grounding failures and apply fixes to the .agent file
  • Run an OWASP LLM Top 10 style security assessment against a live agent and receive a letter grade

Sources

Back to list