Skip to main content

CLI Usage Guide

Complete reference for all Apex AI Brain commands.

Command Syntax​

sfai <command> [options] [arguments]

Foundation Commands​

/init​

Initialize a new Salesforce project context.

sfai /init

What it does:

  • Scans your project structure
  • Extracts Apex classes, LWC components, metadata
  • Builds a context graph
  • Stores configuration in .sfai/ folder

Output:

✓ Project Initialization Complete
  ├─ Project Type: Salesforce DX
  ├─ Files Scanned: 2,847
  ├─ Apex Classes: 42
  ├─ LWC Components: 8
  ├─ Custom Objects: 15
  └─ Context Size: 2.3 MB

/status​

Show current project status and configuration.

sfai /status

Output:

Project Status
==============
Name:          My Salesforce Org
Type:          DX Project
Initialized:   Yes
Last Scanned:  2 hours ago
Complexity:    medium

Agents Ready:  16/16
Context Loaded: Yes

File Statistics:
  - Apex Classes: 42
  - LWC Components: 8
  - Aura Components: 2
  - Custom Objects: 15
  - Workflows: 23

/help​

Display all available commands.

sfai /help

Or get help for a specific command:

sfai /help /plan
sfai /help /spawn

Development Commands​

/plan​

Create a structured plan for a feature request.

sfai /plan "Feature description"

Examples:

sfai /plan "Create customer portal with LWC"
sfai /plan "Add lead scoring with Apex batch job"
sfai /plan "Build multi-select picklist with filtering"

Output: Structured JSON plan with:

  • Task breakdown
  • Agent assignments
  • Dependency mapping
  • Estimated complexity
  • Suggested execution order

/spawn​

Execute a specific agent for a focused task.

sfai /spawn <agent-name> "<task description>"

Examples:

sfai /spawn apex-architect "Design service layer for lead management"
sfai /spawn lwc-builder "Create form component with validation"
sfai /spawn soql-optimizer "Optimize account queries for dashboard"
sfai /spawn test-writer "Write tests for User_Management service"

Agent Names Available:

  • apex-architect - Apex class design & patterns
  • lwc-builder - Lightning Web Component development
  • apex-reviewer - Code review & quality checks
  • test-writer - Apex test class generation
  • soql-optimizer - Query optimization
  • security-agent - Security & compliance checks
  • schema-analyst - Data model design
  • flow-advisor - Flow creation & optimization
  • metadata-manager - Metadata configuration
  • planner - Feature planning
  • git-agent - Git commits & PRs
  • deployment-agent - Deployment strategies
  • debug-agent - Debugging & troubleshooting
  • diff-reviewer - Git diff analysis
  • kanban-agent - Task management
  • memory-agent - Memory management

Agent Management Commands​

/agents​

List all available agents with brief descriptions.

sfai /agents

Output:

Available Agents (16 total)
===========================

🏗  ARCHITECT AGENTS
  ├─ apex-architect          Design Apex structures & patterns
  ├─ lwc-builder             Build Lightning Web Components
  ├─ schema-analyst          Design data models
  └─ flow-advisor            Create & optimize flows

🔍 REVIEWER & QUALITY
  ├─ apex-reviewer           Review Apex code quality
  ├─ security-agent          Check security & compliance
  ├─ test-writer             Write comprehensive tests
  └─ diff-reviewer           Analyze code changes

  OPERATIONS
  ├─ soql-optimizer          Optimize SOQL queries
  ├─ metadata-manager        Manage metadata
  ├─ git-agent               Handle git operations
  ├─ deployment-agent        Plan deployments
  └─ kanban-agent            Manage tasks

🧠 UTILITIES
  ├─ planner                 Break down features
  ├─ debug-agent             Troubleshoot issues
  └─ memory-agent            Manage memory

/agent-spawn​

Alternative syntax for spawning agents.

sfai /agent-spawn <agent-name> "<task>"

Same as /spawn.

/agent-kill​

Stop a running agent.

sfai /agent-kill <agent-id>

/agent-logs​

View agent execution logs.

sfai /agent-logs                    # All logs
sfai /agent-logs <agent-name>      # Specific agent
sfai /agent-logs --tail 50         # Last 50 lines
sfai /agent-logs --filter error    # Only errors

Model & Cost Tracking​

/model​

Manage AI model selection.

sfai /model list              # Show available models
sfai /model set claude        # Set default model
sfai /model info              # Show current settings

Models:

  • claude - Claude (default, most capable)
  • gemini - Google Gemini (fast, cost-effective)
  • gpt4 - OpenAI GPT-4 (experimental)

/tokens​

Check token usage and cost tracking.

sfai /tokens                  # Current session
sfai /tokens --daily         # Today's usage
sfai /tokens --monthly       # This month's usage
sfai /tokens --breakdown     # By model breakdown

Output:

Token Usage Report
==================
Period:    Current Session
Tokens:    45,234
Cost:      $0.68

By Model:
  Claude Opus:       30,000 tokens  ($0.45)
  Gemini Pro:        15,234 tokens  ($0.23)

Monthly Tracking:
  Used:     $87.45
  Budget:   $100.00
  Remaining: $12.55

/cost​

Analyze cost by model and feature.

sfai /cost --by-model        # Cost breakdown by model
sfai /cost --by-agent        # Cost by agent
sfai /cost --projection      # Projected costs

Memory & Context​

/memory​

Manage shared memory storage.

sfai /memory list            # List all stored items
sfai /memory get <key>       # Get specific item
sfai /memory set <key> <value>
sfai /memory clear          # Clear all memory
sfai /memory export         # Export to JSON
sfai /memory import <file>  # Import from JSON

Utility Commands​

/clear​

Clear the terminal screen.

sfai /clear

/exit​

Exit the Apex AI Brain CLI.

sfai /exit

Or use: Ctrl+C

Command Workflow Examples​

Example 1: Building a Complete Feature​

# 1. Plan the feature
sfai /plan "Add customer feedback system"

# 2. Design data model
sfai /spawn schema-analyst "Design feedback custom objects"

# 3. Build Apex layer
sfai /spawn apex-architect "Create FeedbackHandler service with bulkified logic"

# 4. Build LWC layer
sfai /spawn lwc-builder "Create FeedbackForm LWC component"

# 5. Write tests
sfai /spawn test-writer "Write tests for FeedbackHandler"

# 6. Review code
sfai /spawn apex-reviewer "Review FeedbackHandler for best practices"

# 7. Check costs
sfai /tokens --breakdown

# 8. Commit and deploy
sfai /spawn git-agent "Create meaningful commit and PR"

Example 2: Quick Code Review​

sfai /spawn apex-reviewer "Review AccountService class for security issues"
sfai /spawn soql-optimizer "Optimize queries in AccountService"

Example 3: Troubleshooting​

sfai /spawn debug-agent "Why is my LWC component not rendering?"
sfai /spawn security-agent "Check for security vulnerabilities in authentication flow"

Interactive Mode Tips​

The CLI runs in an interactive REPL mode. Tips:

Use arrow keys to navigate command history Type /help for a quick reminder Press Tab for command autocomplete (if enabled) Type keywords to search previous agent outputs Use /memory get to retrieve previous results

Advanced Options​

Environment Variables​

# Set default model globally
export SFAI_DEFAULT_MODEL=claude

# Disable token tracking
export SFAI_DISABLE_TRACKING=true

# Verbose logging
export SFAI_DEBUG=true

Configuration File​

Create sfai.config.json:

{
  "defaultModel": "claude",
  "tokenTracking": {
    "enabled": true,
    "monthlyBudget": 100
  },
  "agents": {
    "timeout": 60000
  }
}

Troubleshooting Commands​

Debug Mode​

DEBUG=true sfai /status

Shows detailed logs and system information.

Check Agent Status​

sfai /agents
sfai /agent-logs --filter error

Verify Installation​

node -v && npm -v
sfai /help
sfai /status

FAQ​

Q: What if an agent times out? A: Check /agent-logs and try with simpler task or use --timeout 120 option.

Q: Can I cancel a running agent? A: Use Ctrl+C or /agent-kill <agent-id>.

Q: How do I export agent outputs? A: Use /memory export to get all results as JSON.

Q: Can I use multiple models in one plan? A: Yes, different agents can use different models automatically.

Q: How do I track costs per feature? A: Check /tokens --breakdown and /memory for feature-specific outputs.