agent-native-architecture - Design prompt-native agent architectures
Designs AI agent systems around prompt-defined outcomes, primitive tools, action parity, shared workspaces, and safe self-modification.
Tags
Updated: 2026-09-30Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Design prompt-native architectures
- Create primitive MCP tools
- Write agent system prompts
- Enable safe self-modification
- Refactor toward prompt-native design
- Inject dynamic runtime context
- Ensure user-agent action parity
- Design shared workspaces
- Test agent-native applications
- Plan mobile execution patterns
- Integrate external APIs
Inputs
- User requirements
- Project architecture context
- Existing codebase
- Available agent capabilities
- Target platform constraints
- External API requirements
Outputs
- Architecture plans
- MCP tool specifications
- System prompt designs
- Refactoring recommendations
- Capability maps
- Testing plans
- Integration guidance