Orcheo Python SDK

The Python SDK offers a strongly typed way to generate Orcheo workflow requests without forcing a specific HTTP client dependency.

Server-side workflow execution

Breaking change: The SDK no longer supports executing workflows locally within the client process. All executions now run on the Orcheo backend. To migrate, deploy your workflow as before and trigger runs via the HTTP API while streaming updates over a WebSocket connection. The new examples/quickstart/sdk_server_trigger.py script shows an end-to-end interaction pattern, including payload construction and streaming result handling.

High-level steps:

Model the workflow graph using the SDK types shown below.
Deploy the workflow to the Orcheo backend.
Trigger runs using OrcheoClient.workflow_trigger_url() or the HttpWorkflowExecutor.
Subscribe to live updates with OrcheoClient.websocket_url() and send the payload returned from client.build_payload(...).

The sections that follow focus on authoring the graph configuration and preparing the payloads required by the server APIs.

Workflow authoring

from pydantic import BaseModel
from orcheo_sdk import OrcheoClient, Workflow, WorkflowNode


class UppercaseConfig(BaseModel):
    prefix: str


class UppercaseNode(WorkflowNode[UppercaseConfig]):
    type_name = "Uppercase"


workflow = Workflow(name="demo")
workflow.add_node(UppercaseNode("upper", UppercaseConfig(prefix="Result: ")))

# Prepare deployment request metadata for the Orcheo backend
client = OrcheoClient(base_url="http://localhost:2025")
request = client.build_deployment_request(workflow)

Multi-node workflows

Edges between nodes are derived from the dependencies you provide when registering each node. Every dependency is converted into an edge in the exported graph, so you only need to describe how data should flow between nodes:

flowchart TD
    S[START] --> first_edge[first]
    S --> second_edge[second]
    first_edge --> combine_edge[combine]
    second_edge --> combine_edge
    combine_edge --> E[END]

    classDef default font-family:monospace,font-size:12px;

workflow = Workflow(name="fan-in")

workflow.add_node(UppercaseNode("first", UppercaseConfig(prefix="A: ")))
workflow.add_node(UppercaseNode("second", UppercaseConfig(prefix="B: ")))
workflow.add_node(
    AppendNode("combine", AppendConfig(suffix="!")),
    depends_on=["first", "second"],
)

graph_config = workflow.to_graph_config()
assert graph_config["edges"] == [
    ("START", "first"),
    ("START", "second"),
    ("combine", "END"),
    ("first", "combine"),
    ("second", "combine"),
]

Note: Workflows should be deployed to the managed Orcheo runtime for execution once you are happy with the authored graph configuration.

Usage

Python Client

from orcheo_sdk import OrcheoClient

client = OrcheoClient(base_url="http://localhost:2025")
trigger_url = client.workflow_trigger_url("example-workflow")
ws_url = client.websocket_url("example-workflow")
payload = client.build_payload(graph_config, inputs={"name": "Ada"})

Refer to examples/quickstart/sdk_server_trigger.py for a complete async implementation that streams updates, handles connection failures gracefully, and shuts down once the workflow run finishes.

Command Line Interface

The SDK includes a comprehensive CLI for managing workflows, nodes, and credentials. All commands support caching and offline mode.

Global Options

Available on all commands:

--profile - Profile name from the CLI config
--api-url - Override the API URL
--service-token - Override the service token
--offline - Use cached data when network calls fail
--cache-ttl - Cache TTL in hours for offline data (default: 24)

Commands Reference

Workflow Management

Command	Description	Key Options
`orcheo workflow list`	List workflows with metadata	`--archived` - Include archived workflows
`orcheo workflow show <id>`	Display workflow details, versions, and runs	-
`orcheo workflow run <id>`	Trigger a workflow execution	`--actor` - Actor name (default: "cli") `--inputs` - JSON inputs payload `--inputs-file` - Path to JSON inputs file
`orcheo workflow delete <id>`	Delete a workflow	`--force` - Skip confirmation prompt
`orcheo workflow upload <file>`	Upload workflow from Python LangGraph script (`.py`)	`--id` - Workflow ID (for updates) `--config` - JSON runnable config payload `--config-file` - JSON runnable config file
`orcheo workflow save-config <id>`	Update version `runnable_config` without creating a version	`--config` / `--config-file` - Runnable config payload `--version` - Target version (latest by default) `--clear` - Clear stored runnable config
`orcheo workflow download <id>`	Download workflow source as Python	`-o, --output` - Output file path `--config-out` - Companion runnable config JSON path

Examples:

# List all workflows including archived ones
orcheo workflow list --archived

# Show workflow details
orcheo workflow show my-workflow-id

# Run a workflow with inputs
orcheo workflow run my-workflow-id --inputs '{"name": "Ada"}'
orcheo workflow run my-workflow-id --inputs-file inputs.json

# Upload a workflow from Python file
orcheo workflow upload workflow.py

# Update existing workflow
orcheo workflow upload workflow.py --id my-workflow-id

# Upload with runnable config defaults (stored per version)
orcheo workflow upload workflow.py --config '{"tags": ["support"], "max_concurrency": 4}'

# Save runnable config only (no new version created)
orcheo workflow save-config my-workflow-id --config '{"tags": ["support"]}'

# Download workflow source as Python
orcheo workflow download my-workflow-id -o workflow.py

# Download workflow source plus stored runnable config defaults
orcheo workflow download my-workflow-id -o workflow.py --config-out workflow.config.json

# Delete a workflow
orcheo workflow delete my-workflow-id --force

Avoid putting secrets in runnable configs; use environment variables or credential vaults instead.

Node Management

Command	Description	Key Options
`orcheo node list`	List registered nodes	`--tag` - Filter by category keyword
`orcheo node show <name>`	Display node metadata and schema	-

Examples:

# List all nodes
orcheo node list

# Filter nodes by tag
orcheo node list --tag ai

# Show node details
orcheo node show AINode

Credential Management

Command	Description	Key Options
`orcheo credential list`	List accessible credentials	`--workflow-id` - Filter by workflow
`orcheo credential create <name>`	Create a new credential	`--provider` - Provider identifier (required) `--secret` - Secret value (required) `--actor` - Actor name (default: "cli") `--access` - Access level: private/shared/public `--workflow-id` - Workflow scope `--scope` - Scopes (repeatable) `--kind` - Credential kind
`orcheo credential delete <id>`	Delete a credential	`--workflow-id` - Workflow scope `--force` - Skip confirmation

Examples:

# List all credentials
orcheo credential list

# List credentials for a workflow
orcheo credential list --workflow-id my-workflow-id

# Create a credential
orcheo credential create my-api-key \
  --provider openai \
  --secret sk-xxx \
  --access private

# Provider-specific credentials used for model switching
orcheo credential create openai_api_key --provider openai --secret sk-openai
orcheo credential create deepseek_api_key --provider deepseek --secret sk-deepseek

# Delete a credential
orcheo credential delete cred-id --force

Code Generation

Command	Description	Key Options
`orcheo code scaffold <id>`	Generate Python trigger snippet	`--actor` - Actor name (default: "cli")

Examples:

# Generate Python code to trigger a workflow
orcheo code scaffold my-workflow-id

Configuration

Environment Variables

ORCHEO_PROFILE - Default profile to use
ORCHEO_API_URL - API base URL (default: http://localhost:2025)
ORCHEO_SERVICE_TOKEN - Service authentication token
ORCHEO_CONFIG_DIR - Override config directory (default: ~/.config/orcheo/)
ORCHEO_CACHE_DIR - Override cache directory (default: ~/.cache/orcheo/)

Profile Configuration

Create ~/.config/orcheo/cli.toml:

[profiles.default]
api_url = "http://localhost:2025"
service_token = "your-token-here"

[profiles.production]
api_url = "https://api.orcheo.example.com"
service_token = "prod-token"

Use profiles with --profile:

orcheo --profile production workflow list

Development

uv sync --all-groups
uv run pytest tests/sdk -q

Metadata

Release files for orcheo-sdk 0.35.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for orcheo-sdk 0.35.3
File	Size	Uploaded
orcheo_sdk-0.35.3.tar.gz	103.5 kB	Sep 30, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for orcheo-sdk 0.35.3
File	Interpreter	ABI	Platform	Reset
orcheo_sdk-0.35.3-py3-none-any.whl 136.8 kB Sep 30, 2026	Python 3	none	any	Details

Total release size: 240.3 kB

Release files / orcheo_sdk-0.35.3.tar.gz

Download URL	orcheo_sdk-0.35.3.tar.gz
Size	103.5 kB
Tags	Source
SHA-256 checksum How to use checksums	`06e4c506371e52dab8955602bc0b5e7a6d9fd1aca0215eadfff1803040dfab23`
BLAKE2b-256 checksum How to use checksums	`d00114528d0074b72242216361dc19afde28d627fb20103c9f45189539d9076f`
Upload date	Sep 30, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/AI-Colleagues/orcheo
Publishing commit: github.com/AI-Colleagues/orcheo/tree/35d9b46c7f676890d7e43b9eb39110f6d396cf81

Workflow

Publishing configuration: github.com/AI-Colleagues/orcheo/blob/35d9b46c7f676890d7e43b9eb39110f6d396cf81/.github/workflows/ci.yml
Publishing logs: github.com/AI-Colleagues/orcheo/actions/runs/36698712435/attempts/1

Artifact

Subject: orcheo_sdk-0.35.3.tar.gz
SHA-256 checksum: 06e4c506371e52dab8955602bc0b5e7a6d9fd1aca0215eadfff1803040dfab23 Verifying attestations

Release files / orcheo_sdk-0.35.3-py3-none-any.whl

Download URL	orcheo_sdk-0.35.3-py3-none-any.whl
Size	136.8 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`58b702dba67d7bab998828afaa999dd5793f3c6971ce2c638295a8531105062d`
BLAKE2b-256 checksum How to use checksums	`0a3a95d67ad51da64214b8a3876ea76d0c0fc0dd575a947a48a33dfdfdf072bb`
Upload date	Sep 30, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/AI-Colleagues/orcheo
Publishing commit: github.com/AI-Colleagues/orcheo/tree/35d9b46c7f676890d7e43b9eb39110f6d396cf81

Workflow

Publishing configuration: github.com/AI-Colleagues/orcheo/blob/35d9b46c7f676890d7e43b9eb39110f6d396cf81/.github/workflows/ci.yml
Publishing logs: github.com/AI-Colleagues/orcheo/actions/runs/36698712435/attempts/1

Artifact

Subject: orcheo_sdk-0.35.3-py3-none-any.whl
SHA-256 checksum: 58b702dba67d7bab998828afaa999dd5793f3c6971ce2c638295a8531105062d Verifying attestations