Skip to main content

Context Awareness

How Apex AI Brain maintains deep project understanding.

What is Context?​

Context is the deep understanding of your project that Apex AI Brain builds during initialization.

Project Files
    ↓
Scanner analyzes structure
    ↓
Creates dependency graphs
    ↓
Stores in memory
    ↓
All agents reference context
    ↓
AI understands your architecture

Context Components​

1. Code Structure​

  • Apex classes and their relationships
  • LWC component hierarchy
  • Metadata organizations
  • Custom objects and fields

2. Patterns​

  • Existing design patterns
  • Naming conventions
  • Folder structure
  • Code organization

3. Complexity​

  • Project size classification
  • Feature maturity levels
  • Dependency complexity
  • Integration count

4. Metadata​

  • Salesforce configuration
  • Custom settings
  • Field definitions
  • Security rules

How Context Works​

┌─────────────────────────┐
│  Your Project Files     │
└────────────┬────────────┘
             │
↓────────────────────────↓
Scanner      Analyzer
             │
      ┌──────▼───────┐
      │ Context Graph│
      │ (in Memory)  │
      └──────┬───────┘
             │
      All Agents Read
      and Reference

Context Awareness in Action​

Example 1: Agent Design Reference​

Agent A designs LeadService

public class LeadService { ... }

Agent B automatically knows about LeadService because it:

  • Read the context graph
  • Found LeadService in project
  • Can reference it in new code

Example 2: Pattern Recognition​

The system recognizes:

  • Your FSD structure
  • Your naming conventions
  • Your existing patterns
  • Your dependencies

Generates code matching your style.

Example 3: Complexity-Based Routing​

For a complex feature:

  • Recognizes complexity from context
  • Routes to powerful agents (Opus)
  • Includes more context in prompts

For simple task:

  • Uses less context
  • Routes to efficient agents (Gemini)
  • Faster execution

Using Context​

View Current Context​

sfai /status        # See context loaded
sfai /memory get "project.context"    # View details

Reload Context​

When project changes:

sfai /init          # Re-scan project

Reset Context​

Start fresh:

rm -rf .sfai/
sfai /init          # Initialize new context

See Also: