Salesforce Skills for AI Coding Agents

Production-ready Agent Skills for Salesforce development. Generate Apex, Flows, LWC, SOQL, and more — with governor limit awareness, security compliance, and deployment orchestration built in.

Works with Claude Code, Cursor, Codex, Gemini CLI, VS Code Copilot, and 50+ AI tools.

Quick Start

npx skills add Clientell-Ai/salesforce-skills

That's it. All 18 skills are now available in your AI coding agent.

Install Individual Skills

npx skills add Clientell-Ai/salesforce-skills@sf-apex       # Apex code generation & review
npx skills add Clientell-Ai/salesforce-skills@sf-test       # Test class generation
npx skills add Clientell-Ai/salesforce-skills@sf-flow       # Flow generation & PB migration
npx skills add Clientell-Ai/salesforce-skills@sf-lwc        # LWC scaffolding
npx skills add Clientell-Ai/salesforce-skills@sf-soql       # SOQL/SOSL query building
npx skills add Clientell-Ai/salesforce-skills@sf-security   # Security & AppExchange audit
npx skills add Clientell-Ai/salesforce-skills@sf-deploy     # Deployment orchestration & CI/CD
npx skills add Clientell-Ai/salesforce-skills@sf-data       # Data migration & bulk operations
npx skills add Clientell-Ai/salesforce-skills@sf-schema     # Schema & permission management
npx skills add Clientell-Ai/salesforce-skills@sf-debug      # Debug log analysis & troubleshooting
npx skills add Clientell-Ai/salesforce-skills@sf-agentforce # Agentforce agents, topics & actions
npx skills add Clientell-Ai/salesforce-skills@sf-permissions # Permission sets & access auditing
npx skills add Clientell-Ai/salesforce-skills@sf-integration # Named Credentials, OAuth, Platform Events
npx skills add Clientell-Ai/salesforce-skills@sf-docs       # Salesforce documentation navigator
npx skills add Clientell-Ai/salesforce-skills@sf-diagram    # Mermaid diagrams from metadata
npx skills add Clientell-Ai/salesforce-skills@sf-omnistudio # OmniStudio (OmniScripts, FlexCards, IPs)
npx skills add Clientell-Ai/salesforce-skills@sf-find       # Skill discovery

Install globally (all projects)

npx skills add Clientell-Ai/salesforce-skills -g

Manual installation

git clone https://github.com/Clientell-Ai/salesforce-skills.git
cd salesforce-skills
./install.sh /path/to/your/salesforce-project

What Problems Do These Skills Solve?

Skills Catalog

Agents (Claude Code)

Specialized subagents for parallel, focused work:

Compatibility

These skills follow the open Agent Skills specification and work with:

IDEs & Editors: Claude Code, Cursor, VS Code Copilot, Windsurf, Cline, Continue, Augment, Junie, OpenCode, Roo Code

Standalone Agents: OpenHands, Goose, Mux, Amp, Gemini CLI, Codex

Platforms: Databricks, GitHub Copilot, Factory AI, Spring AI

And many more.

Prerequisites

1. Salesforce CLI v2+ — Install

   npm install @salesforce/cli -g
   

2. Authenticated Salesforce org

   sf org login web --alias myOrg
   

3. Any compatible AI coding agent — See full list

Configuration

MCP Server (Optional)

Enable the Salesforce DX MCP server for enhanced org interaction. Edit .claude/.mcp.json:

{
  "mcpServers": {
    "salesforce-dx": {
      "command": "npx",
      "args": ["-y", "@anthropic/salesforce-dx-mcp-server"],
      "env": { "SF_ORG_ALIAS": "myOrg" },
      "disabled": false
    }
  }
}

Project Context

Update CLAUDE.md in your Salesforce project with org-specific conventions:

• Custom object naming patterns
• API version
• Deployment targets
• Team coding standards

Benchmark: Do Skills Actually Help?

Yes. Across 15 tasks, skills improve AI-generated Salesforce code by +108% (10.3 → 21.4 out of 25).

Scored by Claude as judge against a Salesforce-specific rubric across 15 tasks. Full results with per-task breakdown | Run it yourself

Evaluate Yourself

# In Claude Code — run the benchmark interactively
/sf-eval

# Run a single task
/sf-eval apex-trigger-bulk

# Static analysis on your own code (no LLM needed)
bash evals/checks/static-checks.sh MyClass.cls

# Headless benchmark (requires API key)
export ANTHROPIC_API_KEY="sk-ant-..."
bash evals/benchmarks/run-benchmark.sh

Project Structure

salesforce-skills/
├── skills/                  # All skills (canonical location)
│   ├── sf-apex/             # Apex + 3 reference files
│   ├── sf-test/             # Testing + 1 reference file
│   ├── sf-flow/             # Flows + 2 reference files
│   ├── sf-lwc/              # LWC + 1 reference file
│   ├── sf-soql/             # SOQL/SOSL + 1 reference file
│   ├── sf-security/         # Security + 2 reference files
│   ├── sf-deploy/           # Deploy + 1 reference file
│   ├── sf-data/             # Data + 1 reference file
│   ├── sf-schema/           # Schema + 1 reference file
│   ├── sf-debug/            # Debug log analysis & troubleshooting
│   ├── sf-agentforce/       # Agentforce agents, topics & actions
│   ├── sf-permissions/      # Permission sets & access auditing
│   ├── sf-integration/      # Named Credentials, OAuth, Platform Events
│   ├── sf-docs/             # Salesforce documentation navigator
│   ├── sf-diagram/          # Mermaid diagrams from metadata
│   ├── sf-omnistudio/       # OmniStudio (OmniScripts, FlexCards, IPs)
│   ├── sf-find/             # Discovery
│   └── sf-eval/             # Skill evaluation & benchmarking
├── .agents/skills/          # Cross-client symlinks
├── .claude/                 # Claude Code config + agents
├── references/              # Shared reference docs
├── scripts/                 # Helper scripts (--help, --format json)
└── evals/                   # Skill evaluation framework

Contributing

See CONTRIBUTING.md for how to add new skills, improve existing ones, or report issues.

License

Apache-2.0 — see LICENSE.

Built by Clientell.