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?

| Pain Point | Skill | What It Does | |------------|-------|-------------| | Writing test classes is tedious | sf-test | Generates comprehensive tests with @TestSetup, 200-record bulk, permission testing, callout mocks, async testing, Stub API | | Deployment failures (37-60% fail rate) | sf-deploy | Resolves dependencies, diagnoses errors, generates CI/CD pipelines, manages scratch orgs and packages | | AppExchange security review failures | sf-security | Audits CRUD/FLS, SOQL injection, sharing violations, PII exposure, XSS — full AppExchange checklist | | Process Builder migration (support ended Dec 2025) | sf-flow | Converts PB to optimized Flows with bypass logic, consolidation, proper XML generation | | Governor limit violations | sf-apex | Generates bulkified Apex with async decision tables, exception handling, integration patterns | | Complex SOQL/SOSL queries | sf-soql | Builds optimized queries with date literals, geolocation, dynamic SOQL, FIELDS(), FOR UPDATE | | LWC scaffolding complexity | sf-lwc | Full bundles with LDS, navigation, LMS, lifecycle hooks, Jest tests, accessibility | | Data migration headaches | sf-data | Bulk API 2.0, relationship ordering, external ID upsert, file migration, sandbox seeding | | Metadata XML boilerplate | sf-schema | All field types, formulas, rollup summaries, Global Value Sets, record types, permission set groups |

Skills Catalog

| Skill | Description | Invoke | |-------|-------------|--------| | sf-apex | Generate and review Apex classes, triggers, batch jobs with governor limit awareness, CRUD/FLS compliance, async patterns (future/queueable/batch/schedulable), integration patterns, and invocable methods | /sf-apex | | sf-test | Generate comprehensive test classes with @TestSetup, bulk data, permission testing, callout mocks, async testing, Platform Event/CDC testing, Stub API, and REST endpoint testing | /sf-test | | sf-flow | Generate Flow metadata XML, migrate Process Builders to Flows with bypass logic, screen flows, orchestrations, scheduled paths, global variables, and flow test coverage | /sf-flow | | sf-lwc | Scaffold Lightning Web Components with LDS, lifecycle hooks, navigation, Lightning Message Service, shadow/light DOM, Jest tests, and SLDS compliance | /sf-lwc | | sf-soql | Build and optimize SOQL/SOSL queries with date literals, FIELDS(), geolocation, dynamic SOQL, record locking, ALL ROWS, and WITH USER_MODE | /sf-soql | | sf-security | Audit codebases for CRUD/FLS violations, SOQL injection, sharing model issues, custom permission checks, Shield encryption awareness, and AppExchange review readiness | /sf-security | | sf-deploy | Orchestrate deployments with scratch orgs, unlocked/2GP packages, destructive changes, Code Analyzer, multiple auth methods, and CI/CD pipeline generation | /sf-deploy | | sf-data | Data migration with Bulk API 2.0, external ID upserts, relationship ordering, record type mapping, file migration, and sandbox seeding | /sf-data | | sf-schema | Scaffold custom objects with formulas, rollup summaries, geolocation, Global Value Sets, record types, Custom Metadata Types, and permission set groups | /sf-schema | | sf-find | Discover the right Salesforce skill for your task | /sf-find |

Agents (Claude Code)

Specialized subagents for parallel, focused work:

| Agent | Purpose | Model | |-------|---------|-------| | apex-reviewer | Read-only code review for governor limits, bulkification, security | Haiku (fast) | | test-generator | Generate comprehensive test classes from existing code | Sonnet | | security-auditor | Scan entire codebases for security vulnerabilities | Haiku (fast) | | deploy-orchestrator | Multi-step deployment management with error diagnosis | Sonnet |

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).

| Category | Without Skills | With Skills | Improvement | |----------|:---:|:---:|:---:| | Security | 1.5/5 | 4.5/5 | +200% | | Governor Limits | 2.5/5 | 4.4/5 | +76% | | Bulkification | 2.1/5 | 4.3/5 | +105% | | Patterns | 2.1/5 | 4.2/5 | +100% | | Completeness | 2.1/5 | 4.0/5 | +90% | | Total | 10.3/25 (41%) | 21.4/25 (86%) | +108% |

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.