use uv

arch/tools/pyproject.toml

@@ -1,32 +1,35 @@
-[tool.poetry]
+[project]
 name = "archgw"
 version = "0.3.15"
 description = "Python-based CLI tool to manage Arch Gateway."
-authors = ["Katanemo Labs, Inc."]
-packages = [
-    { include = "cli" }
+authors = [
+    { name = "Katanemo Labs, Inc." }
 ]
 readme = "README.md"
+requires-python = ">=3.10"
+dependencies = [
+    "archgw_modelserver>=0.3.15",
+    "click>=8.1.7",
+    "jinja2>=3.1.4",
+    "jsonschema>=4.23.0",
+    "setuptools==75.5.0",
+    "pyyaml>=6.0.2",
+]
 
-[tool.poetry.dependencies]
-python = "^3.10"
-archgw_modelserver = "^0.3.15"
-click = "^8.1.7"
-jinja2 = "^3.1.4"
-jsonschema = "^4.23.0"
-setuptools = "75.5.0"
-pyyaml = "^6.0.2"
-
-[tool.poetry.scripts]
+[project.scripts]
 archgw = "cli.main:main"
 
-[tool.poetry.group.dev.dependencies]
-pytest = "^8.4.1"
+[project.optional-dependencies]
+dev = [
+    "pytest>=8.4.1",
+]
 
 [build-system]
-requires = ["poetry-core>=1.0.0"]
-build-backend = "poetry.core.masonry.api"
+requires = ["hatchling"]
+build-backend = "hatchling.build"
 
+[tool.hatch.build.targets.wheel]
+packages = ["cli"]
 
 [tool.pytest.ini_options]
 addopts = ["-v"]

model_server/README.md

@@ -1,2 +1,30 @@
 # Model Server Package #
 This model server package is a dependency of the Arch intelligent prompt gateway. It should not be used alone. Please refer to the [quickstart-guide](https://github.com/katanemo/arch?tab=readme-ov-file#quickstart) for more details on how to get start with Arch.
+
+## Local development
+
+You can start/stop the local server via the CLI entry point exposed by this package.
+
+Using uv (recommended):
+
+```sh
+uv run model_server --help
+# run in foreground (stays attached until Ctrl+C)
+uv run model_server start --port 51000 --foreground
+# run in background (then stop using the CLI)
+uv run model_server start --port 51000
+uv run model_server stop
+```
+
+Alternative without uv:
+
+```sh
+python -m src.cli --help
+# foreground
+python -m src.cli start --port 51000 --foreground
+# background
+python -m src.cli start --port 51000
+python -m src.cli stop
+```
+
+The FastAPI app lives at `src.main:app` and exposes a health check at `/healthz`.

model_server/pyproject.toml

@@ -33,6 +33,8 @@ setuptools = "75.5.0"
 
 [tool.poetry.scripts]
 archgw_modelserver = "src.cli:main"
+# Convenient alias to match the PEP 621 script name used by `uv`
+model_server = "src.cli:main"
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]
@@ -44,3 +46,46 @@ addopts = ["-v", "-s"]
 retries = 2
 retry_delay = 0.5
 cumulative_timing = false
+
+# ---------------------------------------------------------------------------
+# PEP 621 project metadata for tools like `uv` that expect a [project] table.
+# This co-exists with Poetry's configuration above and is used for local runs.
+# ---------------------------------------------------------------------------
+[project]
+name = "archgw_modelserver"
+version = "0.3.15"
+description = "A model server for serving models"
+readme = "README.md"
+requires-python = ">=3.10,<4.0"
+license = { text = "Apache-2.0" }
+authors = [
+    { name = "Katanemo Labs, Inc", email = "info@katanemo.com" }
+]
+
+dependencies = [
+    "fastapi==0.115.0",
+    "torch==2.6.0",
+    "uvicorn==0.31.0",
+    "transformers>=4.37.0,<5.0.0",
+    "accelerate>=1.0.0,<2.0.0",
+    "pydantic>=2.10.1,<3.0.0",
+    "dateparser",
+    "openai>=1.50.2,<2.0.0",
+    "httpx==0.27.2",
+    "pytest-asyncio",
+    "pytest",
+    "opentelemetry-api>=1.28.0,<2.0.0",
+    "opentelemetry-sdk>=1.28.0,<2.0.0",
+    "opentelemetry-exporter-otlp>=1.28.0,<2.0.0",
+    "opentelemetry-instrumentation-fastapi==0.49b0",
+    "overrides>=7.7.0,<8.0.0",
+    "pytest-retry>=1.6.3,<2.0.0",
+    "pytest-httpserver>=1.1.0,<2.0.0",
+    "setuptools==75.5.0",
+]
+
+[project.scripts]
+# Preferred local CLI name
+model_server = "src.cli:main"
+# Backwards-compatible alias matching Poetry's entry point
+archgw_modelserver = "src.cli:main"