CrewAI

Use mycontext cognitive patterns to power CrewAI agents. The to_crewai() export automatically maps your Context to the role, goal, backstory, and expected_output fields CrewAI needs.

pip install mycontext-ai crewai

Quick Start

from mycontext.templates.free.reasoning import RootCauseAnalyzer
from mycontext.integrations import CrewAIHelper
from crewai import Crew, Task

# Build cognitive context
ctx = RootCauseAnalyzer().build_context(
    problem="Customer churn increased 40% in Q3",
    depth="comprehensive",
)

# Create agent with the cognitive framework
agent = CrewAIHelper.create_agent(ctx, name="diagnostician")

# Create matching task
task = CrewAIHelper.create_task(
    ctx,
    description="Diagnose the root causes of the Q3 churn spike",
    agent=agent,
)

# Run the crew
crew = Crew(agents=[agent], tasks=[task], verbose=True)
result = crew.kickoff()
print(result)

ctx.to_crewai()

The native export maps your Context to CrewAI's required fields:

ctx = RootCauseAnalyzer().build_context(
    problem="Production database is 10x slower",
)
crew_cfg = ctx.to_crewai()

# {
#   "role": "Root Cause Analysis Specialist",
#   "goal": "Conduct systematic root cause analysis...",
#   "backstory": "You are a Root Cause Analysis Specialist...",
#   "context": "## Role\n...\n## Directive\n...",
#   "expected_output": "Output must include: root cause, contributing factors, prevention",
#   "tools": [],
#   "verbose": True
# }

from crewai import Agent, Task

agent = Agent(
    role=crew_cfg["role"],
    goal=crew_cfg["goal"],
    backstory=crew_cfg["backstory"],
    verbose=True,
)

task = Task(
    description=crew_cfg["goal"],
    expected_output=crew_cfg["expected_output"],
    agent=agent,
)

How the mapping works

CrewAI field	Source in Context
role	guidance.role
goal	directive.content
backstory	Full guidance.render() (role + rules + style)
expected_output	Derived from constraints.must_include items
context	Full ctx.assemble()

CrewAIHelper Methods

create_agent(context, name, tools, **kwargs)

agent = CrewAIHelper.create_agent(
    context=ctx,
    name="risk_analyst",
    tools=[search_tool, calculator_tool],
    memory=True,           # CrewAI kwargs pass through
    max_iter=10,
)

create_task(context, description, agent, expected_output, **kwargs)

task = CrewAIHelper.create_task(
    context=ctx,
    description="Conduct a comprehensive risk assessment",
    agent=agent,
    expected_output="Risk matrix with probability, impact, and mitigation for each risk",
)

Multi-Agent Crew

Build a full research + analysis crew with different cognitive patterns powering each agent:

from mycontext.templates.free.reasoning import RootCauseAnalyzer, HypothesisGenerator
from mycontext.templates.free.planning import ScenarioPlanner
from mycontext.templates.free.specialized import RiskAssessor
from mycontext.integrations import CrewAIHelper
from crewai import Crew, Process

# Agent 1: Root cause analyst
rca_ctx = RootCauseAnalyzer().build_context(
    problem="Enterprise churn spiked 40% in Q3",
    depth="comprehensive",
)
analyst = CrewAIHelper.create_agent(rca_ctx, name="root_cause_analyst")

# Agent 2: Hypothesis generator
hyp_ctx = HypothesisGenerator().build_context(
    observation="Churn spike correlates with new pricing rollout",
    domain="SaaS business",
)
scientist = CrewAIHelper.create_agent(hyp_ctx, name="hypothesis_generator")

# Agent 3: Risk assessor
risk_ctx = RiskAssessor().build_context(
    decision="Reverse the pricing change vs. offer discounts",
    depth="thorough",
)
risk_agent = CrewAIHelper.create_agent(risk_ctx, name="risk_assessor")

# Tasks
task1 = CrewAIHelper.create_task(
    rca_ctx,
    description="Diagnose the root causes of Q3 churn spike",
    agent=analyst,
)
task2 = CrewAIHelper.create_task(
    hyp_ctx,
    description="Generate testable hypotheses from the diagnosis",
    agent=scientist,
)
task3 = CrewAIHelper.create_task(
    risk_ctx,
    description="Assess risks of each proposed fix",
    agent=risk_agent,
)

# Sequential crew
crew = Crew(
    agents=[analyst, scientist, risk_agent],
    tasks=[task1, task2, task3],
    process=Process.sequential,
    verbose=True,
)
result = crew.kickoff()

Using with Tools

from crewai.tools import BaseTool
from mycontext.templates.free.analysis import DataAnalyzer
from mycontext.integrations import CrewAIHelper

ctx = DataAnalyzer().build_context(
    data_description="Customer usage metrics and support tickets",
    goal="Identify churn predictors",
)

# Pass tools to the agent
agent = CrewAIHelper.create_agent(
    ctx,
    name="data_analyst",
    tools=[
        database_query_tool,
        csv_reader_tool,
        chart_generator_tool,
    ],
)

API Reference

CrewAIHelper

Method	Returns	Description
create_agent(context, name, tools, **kwargs)	Agent	CrewAI Agent
create_task(context, description, agent, expected_output, **kwargs)	Task	CrewAI Task

ctx.to_crewai()

{
    "role": str,             # From guidance.role
    "goal": str,             # From directive.content
    "backstory": str,        # From guidance.render()
    "context": str,          # Full assembled context
    "expected_output": str,  # Derived from constraints.must_include
    "tools": list,           # Empty — user provides tools
    "verbose": bool,         # True
}