Initial Python SDK with generated API models and codegen tooling

.github/workflows/ci.yml

@@ -0,0 +1,47 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+jobs:
+  check:
+    strategy:
+      matrix:
+        os: [ubuntu-latest, macos-latest, windows-latest]
+        python-version: ["3.14"]
+        include:
+          - os: ubuntu-latest
+            python-version: "3.10"
+    runs-on: ${{ matrix.os }}
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: astral-sh/setup-uv@v6
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+          allow-prereleases: true
+
+      - run: uv sync
+
+      - name: Lint
+        run: |
+          uv run poe lint
+          uv run poe typecheck
+
+      - name: Generate API and check for diff
+        run: |
+          uv run poe generate-api
+          git diff --exit-code
+
+      - name: Test
+        env:
+          # Forks won't have these secrets so e2e tests skip, but pushes
+          # to main will run them.
+          BASETEN_E2E_TEST_API_KEY: ${{ secrets.BASETEN_E2E_TEST_API_KEY }}
+          BASETEN_E2E_TEST_DOMAIN: ${{ secrets.BASETEN_E2E_TEST_DOMAIN }}
+          BASETEN_E2E_TEST_MODEL_ID: ${{ secrets.BASETEN_E2E_TEST_MODEL_ID }}
+        run: uv run poe test

.gitignore

@@ -0,0 +1,4 @@
+.pytest_cache
+.ruff_cache
+.venv
+__pycache__

CONTRIBUTING.md

@@ -0,0 +1,41 @@
+# Contributing
+
+## Setup
+
+```bash
+uv sync
+```
+
+## Tasks
+
+All tasks are run via `uv run poe <task>`:
+
+- `generate-api` - Regenerate API clients and models from OpenAPI specs (pass `--update-specs` to download latest specs first)
+- `format` - Format code and auto-fix lint issues
+- `lint` - Check formatting and lint (fails on issues)
+- `typecheck` - Run type checker
+- `test` - Run tests
+
+## End-to-End Tests
+
+E2e tests in `tests/test_e2e.py` run against a live Baseten environment. They are skipped automatically when `BASETEN_E2E_TEST_API_KEY` is not set.
+
+### Bootstrap
+
+Deploy a minimal test model (requires `truss`):
+
+```bash
+BASETEN_E2E_TEST_API_KEY=... BASETEN_E2E_TEST_DOMAIN=... \
+    uv run --with truss python -m scripts.e2e_test_bootstrap
+```
+
+This prints the model ID to stdout. Wait for the model to be ready before running tests.
+
+### Running
+
+```bash
+BASETEN_E2E_TEST_API_KEY=... \
+BASETEN_E2E_TEST_DOMAIN=... \
+BASETEN_E2E_TEST_MODEL_ID=... \
+    uv run poe test tests/test_e2e.py
+```

README.md

@@ -1,3 +1,27 @@
 # Baseten Python SDK
 
-Under development.
+Python SDK for Baseten.
+
+⚠️ Under active development. Nothing should be considered stable at this time.
+
+## Usage
+
+Current SDK only has barebones client. Here is usage example of the barebones underlying client:
+
+```python
+from baseten.client import ManagementClient
+
+with ManagementClient(api_key="my-api-key") as client:
+    for model in client.api.get_models().models:
+        print(model.name)
+```
+
+Or for async:
+
+```python
+from baseten.client import AsyncManagementClient
+
+async with AsyncManagementClient(api_key="my-api-key") as client:
+    for model in (await client.api.get_models()).models:
+        print(model.name)
+```

baseten/client/__init__.py

@@ -0,0 +1,26 @@
+"""Baseten client library.
+
+Use :class:`ManagementClient` or :class:`AsyncManagementClient` for the
+management API, and :class:`InferenceClient` or :class:`AsyncInferenceClient`
+for the inference API.
+"""
+
+from baseten.client._inference import (
+    AsyncInferenceClient,
+    InferenceClient,
+    InferenceClientOptions,
+)
+from baseten.client._management import (
+    AsyncManagementClient,
+    ManagementClient,
+    ManagementClientOptions,
+)
+
+__all__ = [
+    "AsyncInferenceClient",
+    "AsyncManagementClient",
+    "InferenceClient",
+    "InferenceClientOptions",
+    "ManagementClient",
+    "ManagementClientOptions",
+]