Tech Connector API reference

Choose the reference that matches your task: public Python methods · bundled DCC wrappers and host counts · native vendor API names · Garden scene commands · Ophanim actions · Browse engine and DCC source functions · connected services.

Host counts live in the DCC catalog so this overview does not repeat a potentially stale copy. A reference name, source definition, and registered operation each describe a different level of availability.

The desktop interface and the Python API use the same project context, action graphs, DCC bridges, and execution services. Use the API for scripts, CI, local services, or custom interfaces. A planned action or successful dispatch is not proof that a DCC operation changed an asset; inspect the destination output.

Use Search API and guide in the desktop Help window to find a function by name and jump to its entry. On the website, the searchable API explorer links directly to each host function and includes a separate Python method reference. The index is generated from the current source and should be regenerated when API functions change.

Start with the public Python API

• Headless API: setup, create_api, AI generation, discovery, action graphs, prompt chains, Python calls, DCC calls, results, errors, and complete examples.
• Reasoning runtime API: contextual requests, routing, progress events, runtime tools, and modular feature discovery.
• Connected service API: Slack, Discord, Jira, Confluence, and ClickUp workflows and their approval rules.
• Ophanim image editor tools: desktop image and texture actions, separated from the public Python methods.
• Complete DCC function catalog: generated from the current source, with host, call form, arguments, and source location for every indexed public module-level DCC function. Some names are ambiguous; those entries show the required full target path.
• Native host APIs: Maya commands, OpenMaya, and Unreal Python call patterns, live discovery, and the complete versioned vendor references. Native host functions are separate from bundled wrappers.

The website also has a source function index for engine and DCC modules. It includes module functions and class methods, including internal helpers, for code navigation. Those definitions are not all supported Python API entry points or bridge-callable operations. Host pages separately show registered planner and bridge operations; a registered operation is not proof of a live connection or successful asset edit.

from tech_connector.api import create_api

api = create_api(project_root="C:/project")
for operation in api.surface()["operations"]:
    print(operation["path"], operation["signature"])

for host, functions in api.dcc.catalog()["hosts"].items():
    for function in functions:
        print(host, function["api_call"], function["signature"])

api.surface() is the current public core and namespace inventory. The DCC catalog is dynamic; use api.dcc.catalog() to inspect the functions installed in your build. Use api.features.search(...) and api.features.describe(feature, operation) to inspect an operation before calling it. The checked-in catalog reflects the source at generation time and may differ from a packaged build or an unavailable host bridge.

Garden's small Python namespace routes a larger scene-command registry. Call api.garden.commands() for the current command keys and status, or api.garden.search("bevel", department="modeling") to narrow the list. api.garden.describe("modeling.bevel_edges") returns the registered target types, preferred hosts, maturity, and outstanding gates. Then use api.garden.route(command, target) to inspect the route before execution. api.garden.execute_active(command, ...) needs an open Garden viewer; a route or registry entry alone does not execute a command. api.surface() lists the Garden command catalog as a dynamic surface so the small facade count is not mistaken for the total scene-command count. The website guide includes a searchable Garden command reference. Ophanim's five Python methods cover opening, capture, listing, and invoking 32 allowlisted desktop actions. Those actions require an already open editor and the UI thread; they are not headless image-processing functions.

Kingdom exposes runtime creation, character decisions, graphs, and simulation through direct methods. The project editor has a larger declared contract: api.kingdom.editor_capabilities() groups its operations, api.kingdom.editor_describe("create_material") shows the signature, and api.kingdom.editor_call("create_material", ...) invokes a declared operation. The runtime facade also exposes audit_runtime, qualify_runtime, solve_grounded_animation, and step_vehicle for an existing runtime. Qualification returns evidence for the runtime under test; it does not certify an entire release or platform.

Official host APIs

The DCC function catalog counts bundled Tech Connector wrappers. It does not count the host's own API. With an active bridge, call an official Python function by its full path, for example:

maya = api.dcc.maya.call("maya.cmds.ls", kwargs={"selection": True})
unreal_asset = api.dcc.unreal.call(
    "unreal.EditorAssetLibrary.load_asset", "/Game/Characters/Hero"
)
maya_commands = api.dcc.native_catalog("maya")
unreal_classes = api.dcc.native_catalog("unreal")
asset_methods = api.dcc.native_catalog("unreal", "unreal.EditorAssetLibrary")
vendor_names = api.dcc.reference_search("selection", host="maya", limit=20)
for reference in vendor_names["results"]:
    print(reference["name"], reference["execution"])

For native host objects that need to be used together, run the complete script through the bridge:

result = api.dcc.maya.execute_python("""
import maya.api.OpenMaya as om
selection = om.MGlobal.getActiveSelectionList()
print([selection.getDagPath(i).fullPathName() for i in range(selection.length())])
""")
if result.ok:
    print(result.result["output"])
else:
    print(result.error)

This executes in Maya, so MSelectionList and MDagPath remain in its process for the duration of the call. The response is captured text. A connected, authenticated host bridge and an unlocked Tech Connector license are required. The same call and execute_python choice applies to connected Unreal and other Python host bridges. Search results include an execution hint showing the appropriate route, or documentation_only when there is no general Python route in this build. The hint is a routing guide, not a successful connection or callability check.

native_catalog asks the running host for callable members and can be drilled down by module or class path. It also supports 3ds Max (pymxs), Blender (bpy), Houdini (hou), MotionBuilder (pyfbsdk), and Substance 3D Painter (substance_painter) when their Python bridge is connected. Unity uses C# and is not part of this Python discovery path. Photoshop's UXP bridge remains development-only. No RealFlow connector is present in this build. Calls that need host objects or non-JSON arguments may require a script running inside the host. A catalog entry proves discoverability, not successful execution or compatibility with every host version or plugin.

See Autodesk's Maya Python command help and Epic's Unreal Python API reference for the host's versioned method descriptions.

Other reference areas

• LLM queue management: scheduling, cancellation, and queue telemetry.
• Licensing backend contract: provider-neutral client endpoints and metadata boundaries. This is a contract, not a public payment or entitlement issuance endpoint.
• Engine asset system: asset and editor Python API context.

For desktop controls and workflows, return to the desktop guide.