What problem does it solve?
Botcore design principles codify the fundamental rules that guide architecting modular, secure, and extensible agent tooling, preventing drift and inconsistencies across projects.
Core Features & Use Cases
- CommandResult Everywhere: standardizes outcomes as structured data to enable reliable reasoning and auditing.
- Opt-In Composability: allows projects to grow through independent layers without forcing dependencies.
- Connectors as Capability Boundaries: enforces external access via connectors, providing a clear security contract.
- Constrained Agency: limits what agents can do, improving safety and governance.
- Meta-Tool Pattern: reduces token load by exposing a compact, discoverable tool surface.
- Protocol Over Inheritance: promotes decoupled plugin contracts and easier testing.
- Scoped Memory: enforces memory access boundaries for isolation and predictable sharing.
- Explicit Registration: requires explicit surface area declarations for transparency.
- Defense in Depth: layers security controls to reduce risk.
- Don’t Rebuild the Runtime: builds above the runtime without duplicating capabilities.
Quick Start
Describe a new feature for botcore following the principles outlined here.