Configuration Reference

All configuration options for the M365 Knowledge Worker workload.

Workload Options

Option	Type	Default	Description
workers	integer	25	Number of knowledge workers to simulate (1–300)
department	string	operations	Department type. One of: operations, engineering, sales, hr, finance, executive
duration_hours	integer	null	Duration in hours. null runs indefinitely until stopped
enable_ai_generation	boolean	false	Enable LLM-powered email content generation
email_directive	string	null	Custom directive for AI email generation topics

Setting Options via CLI

Pass options as --config key=value flags:

haymaker deploy m365-knowledge-worker \
  --config workers=50 \
  --config department=sales \
  --config duration_hours=8 \
  --config enable_ai_generation=true \
  --config email_directive="Write about Q4 targets"

YAML Configuration Files

You can also deploy from a YAML config file. Store these in the examples/ directory or pass them directly.

Basic Deployment

# examples/basic-deployment.yaml
workload_name: m365-knowledge-worker
workers: 25
department: engineering
duration_hours: 8

Deploy with:

haymaker deploy -f examples/basic-deployment.yaml

AI Email Deployment

# examples/ai-email-deployment.yaml
workload_name: m365-knowledge-worker
workers: 10
department: sales
duration_hours: 4
enable_ai_generation: true
email_directive: "Write emails about Q4 sales targets and client relationships"

Deploy with:

haymaker deploy -f examples/ai-email-deployment.yaml

Environment Variables

Required – M365 Credentials

These credentials authenticate against the Azure AD application registration that has Microsoft Graph API permissions.

Variable	Description
KW_TENANT_ID	Azure AD tenant ID
KW_APP_ID	Application (client) ID with Graph permissions
KW_CLIENT_SECRET	Client secret for the application

export KW_TENANT_ID="your-tenant-id"
export KW_APP_ID="your-app-client-id"
export KW_CLIENT_SECRET="your-client-secret"  # pragma: allowlist secret

These can also be stored in Azure Key Vault and retrieved by the platform at runtime.

Optional – LLM Provider

Required only when enable_ai_generation is true. The workload uses the agent-haymaker[llm] abstraction layer, which supports multiple providers.

Anthropic

export LLM_PROVIDER=anthropic
export ANTHROPIC_API_KEY="sk-ant-..."  # pragma: allowlist secret

Azure OpenAI

export LLM_PROVIDER=azure_openai
export AZURE_OPENAI_ENDPOINT="https://myresource.openai.azure.com"
export AZURE_OPENAI_DEPLOYMENT="gpt-4"

Install AI dependencies:

pip install "haymaker-m365-workloads[ai]"

If the LLM provider is not configured or the API key is missing, the workload falls back to rotating department-specific email templates automatically. No error is raised – telemetry generation continues normally.

Department Activity Patterns

Each department defines its own activity rates. These are built into the workload and selected by the department configuration option.

Department	Email/hr	Teams/hr	Docs/day	Meetings/day
operations	6	8	4	4
engineering	4	15	6	3
sales	12	10	3	8
hr	10	8	5	5
finance	6	5	8	4
executive	8	5	2	10

Additional pattern parameters (configured per department, not user-facing):

Parameter	Default	Description
activity_variance_percent	30	Random variance applied to activity rates
work_start_hour	8	Start of work hours (UTC)
work_end_hour	17	End of work hours (UTC)

Workload Manifest

The workload is described by workload.yaml at the repository root. This manifest declares the workload name, package entrypoint, required permissions, credentials, and configuration schema. The haymaker workload install command reads this file to register the workload with the platform.

Validation

The workload validates configuration at deploy time:

• workers must be a positive integer, maximum 300
• department must be one of the six valid department names
• enable_ai_generation must be a boolean
• email_directive must be a string (if provided)

Invalid configuration returns a list of error messages before any resources are created.