A full-stack developer agent works across the entire request path: from the database schema through the API layer to the rendered UI. It is useful for small teams or projects where features require coordinated changes at every layer without dedicated specialists.
The role-specific prompt for a full-stack agent captures the handoff points: how the frontend and backend communicate, what contracts are shared, and how migrations and API changes must stay in sync. Without this context, a generalist agent tends to make changes that are internally consistent but break the layer it did not focus on.
What this agent owns
- Implement features end-to-end from database migration through rendered UI
- Keep API contracts consistent between frontend consumption and backend implementation
- Write tests at both the unit and integration level for each layer changed
- Identify and flag cross-cutting concerns that require separate review
- Manage shared types and interfaces that cross the frontend/backend boundary
Recommended model: Claude Sonnet
Full-stack work is broad but each layer is individually straightforward; Sonnet handles the breadth without the cost of Opus.
Example tasks
- Add a new resource from schema to API to list and detail views
- Implement real-time status updates using server-sent events end-to-end
- Extract a shared TypeScript type used in both API responses and UI components
- Build a feature flag system from the config table through to conditional UI rendering
# create an agent from this template, then start it
$ fleet agent create --name fullstack-developer--vendor claude-code --template <template-name>
$ fleet agent start fullstack-developer
Find the exact template name with fleet template list.
