Advanced Workflows

Complex command sequences and context combinations for sophisticated projects

✅ Status: Current - Complex command sequences and context combinations for sophisticated projects.

Advanced Usage Guide: Patterns for complex projects using multiple commands, agents, and careful context management within Gemini Cli conversations.

Overview and Usage Guide

Purpose: Advanced GemBoost patterns for complex, multi-step projects that require careful sequencing of commands and context management.

Important

These are conversation patterns, not executing workflows. All work happens within Gemini Cli based on context provided.

Key Concepts

  • • Command sequences within a conversation
  • • Context layering through multiple agents
  • • Progressive refinement approaches
  • • Project phase management (manual, not automated)

Multi-Context Project Patterns

Full-Stack Development Sequence

# E-commerce platform using multiple contexts
# Step 1: Architecture context
@agent-system-architect "design e-commerce architecture"

# Step 2: Security requirements
@agent-security "define security requirements for payments"

# Step 3: Backend implementation
/gb:implement "API with authentication and payment processing"
# Gemini uses accumulated context from previous steps

# Step 4: Frontend implementation
@agent-frontend-architect "design responsive UI"
/gb:implement "React frontend with TypeScript"

# Step 5: Review
/gb:analyze . --focus quality

# Note: Each step builds context within the conversation
# No actual coordination or parallel execution occurs

Problem-Solving Workflow

# Complex troubleshooting approach
# Step 1: Problem understanding
/gb:troubleshoot "application performance issues"

# Step 2: Expert analysis
@agent-performance-engineer "analyze potential bottlenecks"
@agent-backend-architect "review architecture for issues"

# Step 3: Solution design
/gb:design "performance improvement plan"

# Step 4: Implementation
/gb:implement "performance optimizations"

# Context accumulates but doesn't execute

Complex Project Phases

Project Initialization Pattern

# Starting a new project
# Discovery phase
/gb:brainstorm "project concept"
# Gemini explores requirements

# Planning phase
/gb:design "system architecture"
@agent-system-architect "review and refine"

# Documentation
/gb:document --type architecture
/gb:save "project-plan"
# Creates summary for your records (not persistent storage)

Incremental Development Pattern

# Building features incrementally
# Feature 1: Authentication
/gb:implement "user authentication"
/gb:test --focus security
/gb:document --type api

# Feature 2: User Profiles (builds on auth context)
/gb:implement "user profile management"
/gb:test --focus functionality

# Feature 3: Admin Dashboard (uses previous context)
/gb:implement "admin dashboard"
@agent-frontend-architect "ensure consistency"

# Each feature builds on conversation context

Migration Project Pattern

# Legacy system migration
# Phase 1: Analysis
/gb:load legacy-system/
/gb:analyze . --focus architecture --verbose
# Gemini builds understanding

# Phase 2: Planning
@agent-system-architect "design migration strategy"
/gb:workflow "create migration plan"

# Phase 3: Implementation
/gb:implement "compatibility layer"
/gb:implement "new system components"

# Phase 4: Validation
/gb:test --focus compatibility
/gb:document --type migration

# Manual phases, not automated workflow

Enterprise-Scale Patterns

Large Codebase Analysis

# Systematic analysis of large projects
# Overview
/gb:analyze . --overview
# Get high-level understanding

# Focused analysis by module
/gb:analyze auth-module/ --focus security
/gb:analyze api-module/ --focus quality
/gb:analyze frontend/ --focus performance

# Synthesis
@agent-system-architect "synthesize findings"
/gb:workflow "improvement recommendations"

# Note: Sequential analysis, not parallel

Multi-Technology Projects

# Projects with diverse tech stacks
# Backend (Python)
@agent-python-expert "implement FastAPI backend"
/gb:implement "Python API with async support"

# Frontend (React)
@agent-frontend-architect "implement React frontend"
/gb:implement "TypeScript React application"

# Mobile (React Native)
/gb:implement "React Native mobile app"

# Infrastructure
@agent-devops-architect "design deployment"
/gb:implement "Docker configuration"

# Each technology addressed sequentially

Quality Assurance Workflows

Comprehensive Review Pattern

# Multi-aspect code review
# Quality review
/gb:analyze . --focus quality
@agent-quality-engineer "identify improvements"

# Security review
/gb:analyze . --focus security
@agent-security "check for vulnerabilities"

# Architecture review
@agent-system-architect "evaluate design"

# Performance review
@agent-performance-engineer "suggest optimizations"

# Consolidated improvements
/gb:improve . --fix

# Sequential reviews, not parallel analysis

Testing Strategy Pattern

# Comprehensive testing approach
# Test planning
/gb:design "testing strategy"

# Unit tests
/gb:test --type unit
# Gemini generates unit test code

# Integration tests
/gb:test --type integration
# Gemini generates integration test code

# E2E tests
/gb:test --type e2e
# Gemini suggests E2E test scenarios

# Documentation
/gb:document --type testing

# Test code generation, not execution

Session Management Patterns

Long Project Sessions

# Managing context in long conversations
# Start with context
/gb:load project/

# Work progressively
/gb:implement "feature A"
/gb:implement "feature B"
# Context accumulates

# Create checkpoint
/gb:save "session-checkpoint"
# Creates summary for your notes

# Continue work
/gb:implement "feature C"

# Final summary
/gb:reflect
# Reviews conversation progress

Context Refresh Pattern

# When conversation gets too long
# Save current state
/gb:save "work-complete"
# Copy output for next conversation

# In new conversation:
/gb:load project/
"Previous work: [paste summary]"
# Manually restore context

# Continue work
/gb:implement "next feature"

Important Clarifications

What These Workflows ARE

  • ✅Conversation Patterns: Sequences within a single Gemini conversation
  • ✅Context Building: Progressive accumulation of understanding
  • ✅Command Sequences: Ordered use of commands for better results
  • ✅Manual Phases: User-controlled project progression

What These Workflows ARE NOT

  • ❌Automated Workflows: No automatic execution or orchestration
  • ❌Parallel Processing: Everything is sequential
  • ❌Persistent Sessions: Context lost between conversations
  • ❌Performance Optimization: No code executes to optimize

Best Practices

Conversation Management

  1. Keep Related Work Together: Don't split related tasks across conversations
  2. Build Context Progressively: Start broad, then focus
  3. Document Key Decisions: Use /gb:save for important points
  4. Manage Conversation Length: Start new conversation if too long

Command Sequencing

  1. Logical Order: Analysis → Design → Implementation → Testing
  2. Context Accumulation: Later commands benefit from earlier context
  3. Appropriate Depth: Match analysis depth to task complexity
  4. Clear Scope: Focus commands on specific areas

Agent Usage

  1. Strategic Activation: Use agents for specific expertise
  2. Avoid Overload: Too many agents can dilute focus
  3. Manual Control: Use @agent- for precise control
  4. Context Layering: Add agents in logical order

Summary

Advanced workflows in GemBoost are sophisticated conversation patterns that build context progressively within a single Gemini Cli session. They help generate better outputs through careful command sequencing and context management, but do not involve any actual workflow execution, parallel processing, or automation. Success comes from understanding how to layer context effectively within Gemini's conversation scope.