Capability Framework

Taxonomy

BASIS defines a hierarchical capability taxonomy:

capabilities/
├── data/
│   ├── read_public
│   ├── read_user
│   ├── read_sensitive
│   ├── write_user
│   ├── write_system
│   ├── delete
│   └── export
│
├── communication/
│   ├── generate_text
│   ├── send_internal
│   ├── send_external
│   └── publish
│
├── execution/
│   ├── compute
│   ├── schedule
│   ├── invoke_api
│   └── spawn_agent
│
├── financial/
│   ├── view_balance
│   ├── initiate_payment
│   ├── approve_payment
│   └── transfer_funds
│
└── administrative/
    ├── manage_users
    ├── manage_permissions
    └── configure_system

Requirements

REQ-CAP-001: Agents MUST declare all capabilities they may exercise.

REQ-CAP-002: Agents MUST NOT exercise undeclared capabilities.

REQ-CAP-003: Capability declarations MUST be specific, not wildcard.

Trust Thresholds

Capability	Minimum Trust
data/read_public	100
data/read_user	300
communication/send_internal	300
communication/send_external	500
data/read_sensitive	700
financial/approve_payment	700
execution/spawn_agent	900