Skip to main content

SO1 Agent Catalog

The SO1 Automation Factory includes 20 specialized AI agents organized into 7 domains. Each agent is purpose-built for specific tasks within the SO1 ecosystem.

Overview

Total Agents

20 specialized agents

Domains

7 functional domains

Validation Status

225/225 checks passed

FORGE Compliant

100% stage-gate compliant

Agent Domains

DomainAgentsDescription
Orchestration2Multi-agent coordination and gate enforcement
Automation3n8n workflows, webhooks, scheduling
Engineering3TypeScript code (Hono, Next.js, shared)
DevOps3CI/CD, Railway deployment, auditing
Documentation3Mintlify docs, API specs, runbooks
Prompts3Prompt optimization and engineering
Incident3Incident response and analysis

Domain Details

Purpose: Coordinate multiple agents and enforce stage gates
AgentStagePurpose
Factory Orchestrator0Task decomposition and agent assignment
FORGE GatekeeperCross-cuttingGate validation and VTM production
Start with the Factory Orchestrator for complex multi-agent tasks
Purpose: Design and configure n8n workflow automation
AgentStagePurpose
Workflow Architect2Complete n8n workflow design
Webhook Engineer3Webhook integrations (GitHub, Slack)
Schedule Optimizer3Cron and schedule patterns
Typical Output: n8n workflow JSON conforming to schema
Purpose: Generate TypeScript code for SO1 platform
AgentStagePurpose
Hono Backend3API routes, middleware, Zod validation
Next.js Frontend3React components, App Router patterns
TypeScript Shared2Zod schemas, shared types, contracts
Repositories: so1-control-plane-api, so1-console, so1-shared
Purpose: CI/CD pipelines and infrastructure automation
AgentStagePurpose
Railway Deployer4Railway service configurations
GitHub Actions3CI/CD workflow YAML
Pipeline Auditor5Security and compliance audits
Deployment Platform: Railway (primary)
Purpose: Create technical documentation and specifications
AgentStagePurpose
Mintlify Author5MDX documentation pages
API Documenter5OpenAPI specifications
Runbook Writer5Operational runbooks
Output Format: MDX, YAML (OpenAPI), Markdown
Purpose: Prompt engineering and Veritas integration
AgentStagePurpose
Prompt Refiner3Optimize and validate prompts
Chain Architect2Multi-step prompt sequences
Fragment Curator2Reusable prompt fragments
Veritas Integration: Both consume and produce prompts
Purpose: Incident response and post-incident analysis
AgentStagePurpose
Incident CommanderCross-cuttingOrchestrate incident response
Triage ResponderCross-cuttingInitial assessment and routing
Postmortem Analyst5Blameless RCA and postmortems
Trigger: P1/P2 incidents, alert escalations

Quick Reference

By Use Case

I want to…Use this agent
Design an n8n workflowWorkflow Architect
Create a webhook integrationWebhook Engineer
Generate API endpointsHono Backend
Build React componentsNext.js Frontend
Deploy to RailwayRailway Deployer
Write documentationMintlify Author
Document an APIAPI Documenter
Create operational runbooksRunbook Writer
Optimize promptsPrompt Refiner
Respond to incidentsIncident Commander
Coordinate multiple tasksFactory Orchestrator

By Output Type

Output TypeAgents
n8n Workflow JSONWorkflow Architect, Webhook Engineer
TypeScript CodeHono Backend, Next.js Frontend, TypeScript Shared
GitHub Actions YAMLGitHub Actions
Railway ConfigRailway Deployer
MDX DocumentationMintlify Author
OpenAPI SpecAPI Documenter
Markdown RunbooksRunbook Writer
Veritas PromptsPrompt Refiner, Chain Architect, Fragment Curator

Agent Architecture

Every agent follows a consistent structure:
1

YAML Frontmatter

Metadata including name, version, domain, FORGE stage, and Veritas integration
2

Identity Section

Role, domain expertise, and FORGE stage assignment
3

Core Capabilities

Primary functions and domain-specific expertise
4

FORGE Gate Compliance

Entry/exit conditions and decision record format
5

Technical Deliverables

Output formats, schemas, and real-world examples
6

Workflow Process

Step-by-step execution flow
7

Veritas Integration

Prompts consumed and produced
8

Error Handling

Common failures and escalation paths

Getting Started

Validation Status

All 20 agents have been validated against the SO1 agent schema:
./scripts/validate-agents.sh

Results: 225 checks passed, 0 failures
Production Ready: All agents are validated and ready for use