cmo-agent

name	cmo-agent
description	Expert knowledge for working with CMOAgent — a dual-purpose perpetual agent that maps both Marketing and Leadership purposes to separate LoRA adapters. Covers execute_with_purpose(), get_adapter_for_purpose(), get_status(), the purpose-to-adapter mapping pattern, adapter switching internals, and integration with LeadershipAgent and PurposeDrivenAgent.

CMO Agent Skill

Overview

CMOAgent is a dual-purpose agent that executes tasks through either the Marketing or Leadership LoRA adapter based on the task type.

from cmo_agent import CMOAgent

cmo = CMOAgent(agent_id="cmo-001")
await cmo.initialize()

Key Concepts

Dual Purpose Mapping

cmo.purpose_adapter_mapping
# {"marketing": "marketing", "leadership": "leadership"}

cmo.get_adapter_for_purpose("marketing")   # "marketing"
cmo.get_adapter_for_purpose("leadership")  # "leadership"
cmo.get_adapter_for_purpose("finance")     # ValueError!

execute_with_purpose()

The core CMO capability — execute a task using a specific adapter:

# Marketing task (brand, campaigns, demand gen, …)
result = await cmo.execute_with_purpose(
    {"type": "brand_audit", "data": {"brand": "Acme"}},
    purpose_type="marketing",
)
# result["purpose_type"] == "marketing"
# result["adapter_used"] == "marketing"

# Leadership task (decisions, strategy, governance, …)
result = await cmo.execute_with_purpose(
    {"type": "budget_review"},
    purpose_type="leadership",
)

Important: adapter_name is always restored after execution, even on error.

Default Purpose Type

# Default is "marketing"
result = await cmo.execute_with_purpose({"type": "campaign"})
# result["purpose_type"] == "marketing"

Custom Adapter Names

cmo = CMOAgent(
    agent_id="cmo-001",
    marketing_adapter_name="brand-v2",
    leadership_adapter_name="exec-leadership-v2",
)

get_status()

Returns full dual-purpose status:

status = await cmo.get_status()
status["agent_type"]    # "CMOAgent"
status["primary_adapter"]  # "marketing" (always restores to this)
status["purposes"]["marketing"]["adapter"]   # "marketing"
status["purposes"]["leadership"]["adapter"]  # "leadership"
status["purpose_adapter_mapping"]  # {"marketing": "marketing", "leadership": "leadership"}

get_agent_type()

cmo.get_agent_type()  # ["marketing", "leadership"]

Inheritance Chain

PurposeDrivenAgent → LeadershipAgent → CMOAgent

All LeadershipAgent methods (make_decision, get_decision_history, …) and all PurposeDrivenAgent methods (events, goals, MCP, …) are available.

Common Patterns

# Leadership decision using CMO
decision = await cmo.make_decision(
    context={"campaign_budget": 500_000},
    mode="autonomous",
)

# Marketing event handling
await cmo.subscribe_to_event("campaign_update", handler)
await cmo.handle_event({"type": "campaign_update", "data": {}})

Installation

pip install cmo-agent