[POC] Add Data Designer recipe discovery

packages/data-designer-config/src/data_designer/config/__init__.py

@@ -82,6 +82,7 @@
         UniformSamplerParams,
         UUIDSamplerParams,
     )
+    from data_designer.config.script_params import DataDesignerScriptParams  # noqa: F401
     from data_designer.config.seed import (  # noqa: F401
         IndexRange,
         PartitionBlock,
@@ -204,6 +205,8 @@
     "PartitionBlock": (_MOD_SEED, "PartitionBlock"),
     "SamplingStrategy": (_MOD_SEED, "SamplingStrategy"),
     "SeedConfig": (_MOD_SEED, "SeedConfig"),
+    # script params
+    "DataDesignerScriptParams": (f"{_MOD_BASE}.script_params", "DataDesignerScriptParams"),
     # seed_source
     "DataFrameSeedSource": (f"{_MOD_BASE}.seed_source_dataframe", "DataFrameSeedSource"),
     "AgentRolloutFormat": (_MOD_SEED_SOURCE, "AgentRolloutFormat"),

packages/data-designer-config/src/data_designer/config/script_params.py

@@ -0,0 +1,17 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+
+@dataclass(frozen=True, slots=True)
+class DataDesignerScriptParams:
+    """Runtime parameters forwarded to Python config workflows.
+
+    Attributes:
+        argv: Raw workflow arguments passed after the CLI ``--`` separator.
+    """
+
+    argv: tuple[str, ...] = ()

packages/data-designer/src/data_designer/cli/commands/create.py

@@ -3,21 +3,36 @@
 
 from __future__ import annotations
 
+from typing import Annotated
+
 import click
 import typer
 
+from data_designer.cli.commands.generation_args import resolve_generation_config_target
 from data_designer.cli.controllers.generation_controller import GenerationController
 from data_designer.config.utils.constants import DEFAULT_NUM_RECORDS
 from data_designer.interface.results import SUPPORTED_EXPORT_FORMATS
 
 
 def create_command(
-    config_source: str = typer.Argument(
-        help=(
-            "Path or URL to a config file (.yaml/.yml/.json), or a local Python module (.py)"
-            " that defines a load_config_builder() function."
+    workflow_args: Annotated[
+        list[str] | None,
+        typer.Argument(
+            metavar="[CONFIG_SOURCE] [-- WORKFLOW_ARGS]",
+            help=(
+                "Path or URL to a config file (.yaml/.yml/.json), or a local Python module (.py)"
+                " that defines a load_config_builder() function. Extra arguments after '--' are forwarded to Python"
+                " workflows."
+            ),
         ),
-    ),
+    ] = None,
+    recipe: Annotated[
+        str | None,
+        typer.Option(
+            "--recipe",
+            help="Name of an installed Data Designer recipe to run instead of a config source.",
+        ),
+    ] = None,
     num_records: int = typer.Option(
         DEFAULT_NUM_RECORDS,
         "--num-records",
@@ -67,9 +82,12 @@ def create_command(
         # Create from a Python module with custom output path
         data-designer create my_config.py --artifact-path /path/to/output
     """
+    target = resolve_generation_config_target(workflow_args, recipe)
     controller = GenerationController()
     controller.run_create(
-        config_source=config_source,
+        config_source=target.config_source,
+        recipe=target.recipe,
+        workflow_args=target.workflow_args,
         num_records=num_records,
         dataset_name=dataset_name,
         artifact_path=artifact_path,

packages/data-designer/src/data_designer/cli/commands/generation_args.py

@@ -0,0 +1,33 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+import click
+
+
+@dataclass(frozen=True)
+class GenerationConfigTarget:
+    """Resolved config target for create, preview, and validate commands."""
+
+    config_source: str | None
+    recipe: str | None
+    workflow_args: tuple[str, ...]
+
+
+def resolve_generation_config_target(
+    raw_args: list[str] | None,
+    recipe: str | None,
+) -> GenerationConfigTarget:
+    """Split variadic CLI args into a config source or recipe plus workflow args."""
+    args = tuple(raw_args or ())
+    if recipe is not None:
+        return GenerationConfigTarget(config_source=None, recipe=recipe, workflow_args=args)
+
+    if not args:
+        raise click.UsageError("Missing argument 'CONFIG_SOURCE'. Provide a config source or use --recipe.")
+
+    config_source, *workflow_args = args
+    return GenerationConfigTarget(config_source=config_source, recipe=None, workflow_args=tuple(workflow_args))

packages/data-designer/src/data_designer/cli/commands/preview.py

@@ -3,20 +3,35 @@
 
 from __future__ import annotations
 
+from typing import Annotated
+
 import click
 import typer
 
+from data_designer.cli.commands.generation_args import resolve_generation_config_target
 from data_designer.cli.controllers.generation_controller import GenerationController
 from data_designer.config.utils.constants import DEFAULT_DISPLAY_WIDTH, DEFAULT_NUM_RECORDS
 
 
 def preview_command(
-    config_source: str = typer.Argument(
-        help=(
-            "Path or URL to a config file (.yaml/.yml/.json), or a local Python module (.py)"
-            " that defines a load_config_builder() function."
+    workflow_args: Annotated[
+        list[str] | None,
+        typer.Argument(
+            metavar="[CONFIG_SOURCE] [-- WORKFLOW_ARGS]",
+            help=(
+                "Path or URL to a config file (.yaml/.yml/.json), or a local Python module (.py)"
+                " that defines a load_config_builder() function. Extra arguments after '--' are forwarded to Python"
+                " workflows."
+            ),
         ),
-    ),
+    ] = None,
+    recipe: Annotated[
+        str | None,
+        typer.Option(
+            "--recipe",
+            help="Name of an installed Data Designer recipe to run instead of a config source.",
+        ),
+    ] = None,
     num_records: int = typer.Option(
         DEFAULT_NUM_RECORDS,
         "--num-records",
@@ -54,9 +69,12 @@ def preview_command(
     ),
 ) -> None:
     """Generate a preview dataset for fast iteration on your configuration."""
+    target = resolve_generation_config_target(workflow_args, recipe)
     controller = GenerationController()
     controller.run_preview(
-        config_source=config_source,
+        config_source=target.config_source,
+        recipe=target.recipe,
+        workflow_args=target.workflow_args,
         num_records=num_records,
         non_interactive=non_interactive,
         save_results=save_results,

packages/data-designer/src/data_designer/cli/commands/recipes.py

@@ -0,0 +1,108 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+from typing import Annotated
+
+import click
+import typer
+
+from data_designer.cli.ui import console, print_error, print_header
+from data_designer.cli.utils.config_loader import ConfigLoadError
+from data_designer.cli.utils.recipe_loader import get_recipe_details, get_recipe_help_text, list_recipes
+
+OutputFormat = click.Choice(["text", "json"], case_sensitive=False)
+
+
+def list_command(
+    output: Annotated[
+        str,
+        typer.Option(
+            "--output",
+            "-o",
+            click_type=OutputFormat,
+            help="Output format.",
+        ),
+    ] = "text",
+) -> None:
+    """List installed Data Designer recipes."""
+    recipes = list_recipes()
+    if output == "json":
+        console.print_json(
+            data={
+                "schema_version": "data-designer.recipes.list.v1",
+                "recipes": [recipe.to_dict() for recipe in recipes],
+            }
+        )
+        return
+
+    print_header("Installed Data Designer Recipes")
+    if not recipes:
+        console.print("  No recipes found.")
+        return
+
+    for recipe in recipes:
+        package = recipe.package or "unknown package"
+        version = f" {recipe.version}" if recipe.version is not None else ""
+        console.print(f"  [bold]{recipe.name}[/bold] ({package}{version})")
+
+
+def show_command(
+    recipe_name: Annotated[str, typer.Argument(help="Installed recipe name.")],
+    output: Annotated[
+        str,
+        typer.Option(
+            "--output",
+            "-o",
+            click_type=OutputFormat,
+            help="Output format.",
+        ),
+    ] = "text",
+) -> None:
+    """Show metadata for an installed Data Designer recipe."""
+    try:
+        details = get_recipe_details(recipe_name)
+    except ConfigLoadError as exc:
+        print_error(str(exc))
+        raise typer.Exit(code=1) from exc
+
+    if output == "json":
+        console.print_json(
+            data={
+                "schema_version": "data-designer.recipes.show.v1",
+                "recipe": details.to_dict(),
+            }
+        )
+        return
+
+    print_header(f"Recipe: {details.summary.name}")
+    console.print(f"  Entry point: [bold]{details.summary.entry_point}[/bold]")
+    if details.summary.package is not None:
+        version = f" {details.summary.version}" if details.summary.version is not None else ""
+        console.print(f"  Package: [bold]{details.summary.package}{version}[/bold]")
+    if details.description is not None:
+        console.print(f"  Description: {details.description}")
+
+    if not details.arguments:
+        console.print("  Structured argument metadata: unavailable")
+        return
+
+    console.print("  Arguments:")
+    for argument in details.arguments:
+        flags = ", ".join(argument["flags"]) if argument["flags"] else argument["name"]
+        required = " required" if argument["required"] else ""
+        default = "" if argument["default"] is None else f" default={argument['default']!r}"
+        help_text = "" if argument["help"] is None else f" — {argument['help']}"
+        console.print(f"    [bold]{flags}[/bold]{required}{default}{help_text}")
+
+
+def help_command(
+    recipe_name: Annotated[str, typer.Argument(help="Installed recipe name.")],
+) -> None:
+    """Show recipe-specific workflow argument help."""
+    try:
+        console.print(get_recipe_help_text(recipe_name), markup=False)
+    except ConfigLoadError as exc:
+        print_error(str(exc))
+        raise typer.Exit(code=1) from exc

packages/data-designer/src/data_designer/cli/commands/validate.py

@@ -3,18 +3,33 @@
 
 from __future__ import annotations
 
+from typing import Annotated
+
 import typer
 
+from data_designer.cli.commands.generation_args import resolve_generation_config_target
 from data_designer.cli.controllers.generation_controller import GenerationController
 
 
 def validate_command(
-    config_source: str = typer.Argument(
-        help=(
-            "Path or URL to a config file (.yaml/.yml/.json), or a local Python module (.py)"
-            " that defines a load_config_builder() function."
+    workflow_args: Annotated[
+        list[str] | None,
+        typer.Argument(
+            metavar="[CONFIG_SOURCE] [-- WORKFLOW_ARGS]",
+            help=(
+                "Path or URL to a config file (.yaml/.yml/.json), or a local Python module (.py)"
+                " that defines a load_config_builder() function. Extra arguments after '--' are forwarded to Python"
+                " workflows."
+            ),
+        ),
+    ] = None,
+    recipe: Annotated[
+        str | None,
+        typer.Option(
+            "--recipe",
+            help="Name of an installed Data Designer recipe to validate instead of a config source.",
         ),
-    ),
+    ] = None,
 ) -> None:
     """Validate a Data Designer configuration.
 
@@ -31,5 +46,8 @@ def validate_command(
         # Validate a Python module
         data-designer validate my_config.py
     """
+    target = resolve_generation_config_target(workflow_args, recipe)
     controller = GenerationController()
-    controller.run_validate(config_source=config_source)
+    controller.run_validate(
+        config_source=target.config_source, recipe=target.recipe, workflow_args=target.workflow_args
+    )