Cyborg deployment map¶
What's on this page
- Kubernetes-oriented namespace layout for agent personas
- Security levels and human approval gates
- Family Office invocation restrictions
- Mapping to
cyborgs/*.yaml
What this enables
- Safe multi-agent operation alongside humans
- Clear blast-radius boundaries for automation
Namespaces (logical)¶
| Namespace | Personas |
|---|---|
cyborg-executive |
EXEC, BOARD assistants |
cyborg-development |
SWEN*, PROD engineering-adjacent |
cyborg-sre |
SREL*, platform reliability |
cyborg-security |
AppSec, GRC, SOC personas |
cyborg-gtm |
SALE, MKTG, COMM, CSM |
cyborg-ga |
FINC, PEOP, LEGL, OPS, REAL* |
cyborg-data |
DATA* |
cyborg-customer |
CUST* |
cyborg-family-office |
PERS (restricted*) |
cyborg-squad |
SQAD* orchestrators |
Reliability & latency¶
Default from ported catalog: FOUR_NINES reliability tier and latency budget unless overridden. Heavy batch personas may use THREE_NINES.
Security levels¶
| Level | Human approval | Examples |
|---|---|---|
| CRITICAL | Always for side effects | CEO, CLO, FO, finance disbursement, security break-glass |
| HIGH | Required for prod/security changes | SRE, Security, Counsel |
| MEDIUM | Spot checks / sampling | IC builders, analysts |
| LOW | Informational agents | Read-only research assistants |
Family Office rules¶
- Deploy only into
cyborg-family-office allowed_invokerstypicallyEXEC0001(CEO) andPERS0001(FO Director)- Separate audit trail from corporate GTM agents
- Never train public models on FO conversation content
Machine specs live in repo-root cyborgs/. Human overview: Cyborgs.