Each agent is a specialized persona for a specific development task. Fork into your project's
ai-dev/agents/and customize with project context.
| Agent | File | Primary Use |
|---|---|---|
| Solutions Architect | architect.md |
System design, module interfaces, structural review |
| Python Expert | python-expert.md |
Business logic, ArcPy, data pipelines, idiomatic Python |
| C# / .NET Expert | csharp-expert.md |
ArcGIS Pro SDK, WPF/MVVM, .NET patterns |
| TypeScript Expert | typescript-expert.md |
React/Next.js, Three.js, Vite, web apps |
| Data & Database Expert | data-expert.md |
SQL, ETL, geodatabases, data modeling |
| QA Reviewer | qa-reviewer.md |
Testing, edge cases, code review findings |
| DevOps Expert | devops-expert.md |
Deployment, CI/CD, logging, scheduling |
| Technical Writer | technical-writer.md |
Docs, SOPs, README, runbooks, metadata |
| Frontend & UI Expert | frontend-expert.md |
React UI, CSS, accessibility, data viz |
| Security Reviewer | security-reviewer.md |
Threat modeling, credential audits, input validation |
| GIS Domain Expert | gis-expert.md |
Enterprise GIS, spatial analysis, geodatabase design |
| Refactor Coach | refactor-coach.md |
Legacy modernization, code smells, migration strategy |
Reference from CLAUDE.md:
Read ai-dev/agents/architect.md for your role definition.
Or in a Claude Code / chat prompt:
Read ai-dev/agents/python-expert.md, then implement the data sync module.
When forking an agent into a project:
- Replace
[PROJECT_NAME]with actual project name - Add project-specific patterns to the Patterns section
- Add project-specific anti-patterns
- Update the review checklist with project conventions
- Cross-reference with
ai-dev/guardrails/in the preamble