Engineering Agents
Relevant source files
The following files were used as context for generating this wiki page:
Engineering Agents in the SuperGemini Framework are specialized AI domain experts designed to provide focused technical expertise during the software development lifecycle. These agents do not operate as separate processes but rather through a "Persona Mode" where the Gemini CLI embodies a specific role—such as a Security Engineer or DevOps Architect—based on instruction injection from Markdown configuration files.
Sources: Docs/User-Guide/agents.md:14-18, CLAUDE.md:148-151
These agents facilitate systematic development by providing automated expert coordination, quality gates, and specialized problem-solving methodologies. They are activated automatically through a sophisticated routing system that analyzes user keywords, file types, and task complexity to engage the most relevant specialists for any given request.
Sources: Docs/User-Guide/agents.md:20-24, CLAUDE.md:173-177
Agent Activation and Orchestration Architecture
The framework employs an intelligent routing system to manage the "SuperGemini Agent Team." This system ensures that the correct specialists are engaged without manual user intervention, supporting multi-agent collaboration for complex tasks.
Activation Logic and Priority
The orchestration layer follows a specific priority hierarchy to determine which engineering agents should be activated:
- Keywords: Direct domain terminology (e.g., "auth", "latency") triggers primary agents.
- File Types: Extensions (e.g.,
.py,Dockerfile) activate language or platform specialists. - Complexity: Multi-step tasks engage coordination agents or system architects.
- Context: Related concepts trigger complementary quality or security agents.
Sources: Docs/User-Guide/agents.md:28-33, Docs/User-Guide/agents.md:650-664
Request Routing Flow
The following diagram illustrates how a user request is processed and routed to the specialized Engineering Agents.
The Agent Router analyzes intent and complexity to select domain specialists, who then coordinate with MCP servers for enhanced capabilities.
Sources: Docs/Getting-Started/quick-start.md:38-44, Docs/User-Guide/agents.md:37-43
Core Engineering Agent Specialists
The framework identifies several primary "Quality & Analysis" and "Architecture" agents that handle the bulk of technical engineering tasks.
Security Engineer
The Security Engineer is a critical agent with "veto authority" in the system, focusing on threat modeling, vulnerability prevention, and compliance.
- Auto-Activation Keywords: "security", "auth", "vulnerability", "encryption", "compliance", "OWASP".
- Capabilities: Secure authentication design (OAuth, JWT), data encryption at rest/transit, and vulnerability assessments.
- Integration: Works closely with the Backend Architect to ensure API security.
Sources: Docs/User-Guide/agents.md:154-171, SuperGemini/Agents/security-engineer.md:1-20
Performance Engineer
This agent focuses on system optimization, scalability, and resource efficiency.
- Auto-Activation Keywords: "performance", "slow", "optimization", "bottleneck", "latency", "memory".
- Capabilities: Profiling, database query optimization, caching strategies (Redis, CDN), and load testing.
- Thresholds: Often triggered by response times >500ms or high CPU/memory metrics.
Sources: Docs/User-Guide/agents.md:183-200, SuperGemini/Agents/performance-engineer.md:1-18
Quality Engineer
The Quality Engineer ensures comprehensive testing strategy and adherence to quality metrics.
- Auto-Activation Keywords: "test", "QA", "validation", "coverage", "automation".
- Capabilities: Test strategy design (unit, integration, e2e), quality metrics monitoring, and edge case identification.
- Output: Produces test suites and coverage reports.
Sources: Docs/User-Guide/agents.md:225-241, SuperGemini/Agents/quality-engineer.md:1-20
DevOps Architect
Responsible for infrastructure automation and reliable software delivery pipelines.
- Auto-Activation Keywords: "deploy", "CI/CD", "Docker", "Kubernetes", "infrastructure", "pipeline".
- Capabilities: CI/CD pipeline design, container orchestration, Infrastructure as Code (Terraform), and monitoring/logging stacks.
Sources: Docs/User-Guide/agents.md:135-151, SuperGemini/Agents/devops-architect.md:1-20
Collaborative Workflows
Agents rarely work in isolation. For complex engineering tasks, the framework triggers "Coordination Patterns" where multiple agents collaborate.
Coordination Sequence Diagram
This diagram shows the interaction between multiple engineering agents during a secure feature implementation.
Engineering agents interact synchronously to ensure that security and quality requirements are integrated directly into the implementation flow.
Sources: Docs/User-Guide/agents.md:387-393, Docs/Reference/advanced-patterns.md:46-59
Common Engineering Agent Combinations
| Workflow Type | Primary Agents | Secondary/Supporting Agents |
|---|---|---|
| Full-Stack Development | Frontend Architect, Backend Architect | Security Engineer, Quality Engineer |
| System Optimization | Performance Engineer, System Architect | DevOps Architect, Root Cause Analyst |
| Security Audit | Security Engineer, Quality Engineer | Root Cause Analyst, Technical Writer |
| CI/CD Setup | DevOps Architect, Security Engineer | Quality Engineer, Performance Engineer |
Sources: Docs/User-Guide/agents.md:387-405, Docs/User-Guide/agents.md:667-680
Implementation Details
Agents are defined as Markdown instruction files located in SuperGemini/Agents/. These files contain structured behavioral rules that are injected into the Gemini session.
Agent Configuration Structure
Each agent file follows a standard format:
- Triggers: Specific conditions or keywords that invoke the persona.
- Focus Areas: Specific technical domains the agent prioritizes.
- Key Actions: Step-by-step procedures the agent follows (e.g., "Analyze security requirements").
- Outputs: Standardized deliverables like API docs or test plans.
Sources: SuperGemini/Agents/quality-engineer.md:1-25, CLAUDE.md:154-169
Code Snippet: Agent Trigger Definition
Example metadata structure for the Security Engineer:
---
name: security-engineer
description: Application security architecture and vulnerability prevention
category: security
---
# Security Engineer
## Triggers
- Security architecture review and threat modeling requests
- Authentication/Authorization system implementation...
Sources: SuperGemini/Agents/security-engineer.md:1-10
Summary
Engineering Agents represent the core technical intelligence of the SuperGemini Framework. By leveraging specialized personas like the Security, Performance, and Quality Engineers, the framework provides a multi-perspective analysis of software projects. This systematic approach ensures that non-functional requirements—such as security and performance—are considered alongside functional implementation, resulting in higher-quality software architecture and robust delivery pipelines.
Sources: Docs/User-Guide/agents.md:580-595, Docs/Developer-Guide/technical-architecture.md:20-25