Every Operark organization runs the same 15 named runtime agents: 3 Board,
3 L5 Meta, and 9 C-Suite. You invoke them by their roleKey (for example,
AI_CFO).
C-Suite (L4)
These are the executives you interact with most. Each has its own remit, tools,
and decision-making temperament.
| roleKey | Name | Role |
|---|
AI_CEO | Atlas | Chief Executive Officer — strategy, resource allocation |
AI_CFO | Nova | Chief Financial Officer — planning, treasury, compliance |
AI_CRO | Sentinel | Chief Risk Officer — risk, veto authority, governance |
AI_CTO | Forge | Chief Technology Officer — engineering, architecture |
AI_COO | Tempo | Chief Operating Officer — operations, people, process |
AI_CMO | Spark | Chief Marketing Officer — brand, growth |
AI_CPO | Canvas | Chief Product Officer — product, UX, roadmap |
AI_CRevO | Vanguard | Chief Revenue Officer — pipeline, partnerships |
AI_CLO | Shield | Chief Legal Officer — legal, compliance, contracts |
System-level agents that orchestrate, audit, and respond to incidents.
| roleKey | Role |
|---|
L5_NEXUS | Orchestration — routing, workflow, conflict mediation |
L5_PULSE | Quality & governance — auditing, standards |
L5_ORACLE | Incident command — anomaly detection, crisis response |
Board
Independent oversight, activated for strategic, high-value, and governance
matters.
| roleKey | Role |
|---|
BOARD_CHAIR | Chairman — independent oversight |
BOARD_AUDIT | Audit committee — financial & compliance audit |
BOARD_RISK | Risk committee — enterprise risk oversight |
Invoking an agent
Use ask for an advisory answer and decide for a recorded, governed decision:
# Advisory
curl -X POST "$OPERARK_BASE_URL/api/v1/executives/AI_CTO/ask" \
-H "Authorization: Bearer $OPERARK_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "message": "Is a managed Postgres or self-hosted right for us at our scale?" }'
Only the 15 runtime agents and platform entrypoints can be invoked directly.
Requesting an unknown or non-invocable role returns
ROLE_NOT_DIRECTLY_INVOCABLE. Department and execution capabilities run as
skill clusters inside their parent C-Suite agent.
Board and certain L5 agents require a higher tier; a blocked call returns a clear
code (for example, BOARD_REQUIRES_ENTERPRISE) explaining what is needed.