Deployment Configuration
When deploying a workflow, Agent Studio applies a DeploymentConfig that controls runtime behavior — LLM generation parameters, tool configuration, MCP server secrets, and environment variables.
DeploymentConfig
class DeploymentConfig(BaseModel):
generation_config: Dict = {}
tool_config: Dict[str, Dict[str, str]] = {}
mcp_config: Dict[str, Dict[str, str]] = {}
llm_config: Dict = {}
environment: Dict = {}
| Field | Type | Description |
|---|---|---|
generation_config | object | Default LLM generation parameters applied to all language models |
tool_config | object | Tool-specific configuration keyed by tool instance ID |
mcp_config | object | MCP server secrets keyed by MCP instance ID |
llm_config | object | LLM API keys and endpoint overrides |
environment | object | Additional environment variables passed to the deployment |
Default Generation Config
When no overrides are provided, Agent Studio uses these defaults:
{
"do_sample": true,
"temperature": 0.1,
"max_new_tokens": 4096,
"top_p": 1,
"top_k": 50,
"num_beams": 1,
"max_length": null
}
Language Model Configuration
At deployment time, each language model receives connection details:
class Input__LanguageModelConfig(BaseModel):
provider_model: str # Provider-specific model identifier (e.g., "gpt-4o")
model_type: SupportedModelTypes # Provider type enum
api_base: Optional[str] # Custom API endpoint
api_key: Optional[str] # Authentication key
extra_headers: Optional[Dict[str, str]] # Additional HTTP headers
# AWS Bedrock specific
aws_region_name: Optional[str]
aws_access_key_id: Optional[str]
aws_secret_access_key: Optional[str]
aws_session_token: Optional[str]
Supported Model Types
| Type | Description | Required Fields |
|---|---|---|
OPENAI | OpenAI API | api_key |
OPENAI_COMPATIBLE | OpenAI-compatible endpoints | api_base, api_key |
AZURE_OPENAI | Azure OpenAI Service | api_base, api_key |
GEMINI | Google Gemini | api_key |
ANTHROPIC | Anthropic Claude | api_key |
CAII | Cloudera AI Inference | api_base, api_key |
BEDROCK | AWS Bedrock | aws_region_name, aws_access_key_id, aws_secret_access_key |
Deployment Targets
Agent Studio supports multiple deployment target types:
| Target | Description |
|---|---|
workbench_model | CML Model endpoint — the primary deployment target |
langgraph_server | LangGraph Server deployment |
ai_inference | Cloudera AI Inference (planned) |
model_registry | Model Registry deployment (planned) |
Workflow Source Types
Workflows can be deployed from multiple sources:
| Source | Description |
|---|---|
workflow | An existing workflow in the current Agent Studio instance |
workflow_template | A workflow template (instantiates first, then deploys) |
workflow_artifact | A pre-packaged tar.gz artifact |
github | A GitHub repository URL (cloned and packaged by Agent Studio) |
Environment Variables at Runtime
Deployed workflows receive these environment variables:
| Variable | Description |
|---|---|
AGENT_STUDIO_OPS_ENDPOINT | Phoenix observability endpoint for trace export |
AGENT_STUDIO_WORKFLOW_ARTIFACT | Path to the extracted deployment artifact |
AGENT_STUDIO_WORKFLOW_DEPLOYMENT_CONFIG | JSON-encoded DeploymentConfig |
AGENT_STUDIO_MODEL_EXECUTION_DIR | Working directory for model execution |
CDSW_APIV2_KEY | CML API v2 authentication key |
CDSW_PROJECT_ID | CML project identifier |
CREWAI_DISABLE_TELEMETRY | Set to disable CrewAI’s built-in telemetry |