Merge branch 'main' into rename-letsgo-to-go

Author: mattf

.github/CODEOWNERS

@@ -2,4 +2,8 @@
 
 # These owners will be the default owners for everything in
 # the repo. Unless a later match takes precedence,
-* @ashwinb @raghotham @ehhuang @leseb @bbrowning @mattf @franciscojavierarceo @cdoern
+* @ashwinb @raghotham @ehhuang @leseb @bbrowning @mattf @franciscojavierarceo @cdoern @skamenan7
+
+# OpenAPI SDK generation and publishing
+/client-sdks/openapi/ @ashwinb @leseb @bbrowning
+/.github/workflows/openapi-generator-validation.yml @ashwinb @leseb @bbrowning

.github/ISSUE_TEMPLATE/config.yml

@@ -4,6 +4,6 @@ contact_links:
   - name: Have you read the docs?
     url: https://ogx-ai.github.io/docs
     about: Much help can be found in the docs
-  - name: Chat on Slack
-    url: https://join.slack.com/t/ogx-ai/shared_invite/zt-3uyw5bxj9-tSEwsNZncgkGEKbd4dXIpw
-    about: Maybe chatting with the community can help
+  - name: Join the Discord community
+    url: https://discord.gg/bUYRqEvK6
+    about: Ask questions and chat with the community

.github/actions/install-ogx-client/action.yml

@@ -30,9 +30,10 @@ runs:
       id: configure
       shell: bash
       run: |
-        # If sdk_install_url is provided (e.g., from Stainless preview), use it directly
+        # If sdk_install_url is provided, use it directly
         if [ -n "${{ inputs.sdk_install_url }}" ]; then
           echo "Using provided sdk_install_url: ${{ inputs.sdk_install_url }}"
+          echo "generate=false" >> $GITHUB_OUTPUT
           echo "install-after-sync=true" >> $GITHUB_OUTPUT
           echo "install-source=${{ inputs.sdk_install_url }}" >> $GITHUB_OUTPUT
           exit 0
@@ -51,28 +52,60 @@ runs:
 
         echo "Working with branch: $BRANCH"
 
-        # On release branches: install client from the matching release branch in the client repo
-        # On non-release branches: install based on client-version after sync
+        # The ogx-client Python SDK is generated in-repo from client-sdks/openapi.
+        # On release branches and when client-version=latest, generate the SDK from
+        # the current checkout so tests run against the client built from this source.
+        # When client-version=published, use the version resolved by uv sync from PyPI.
         if [[ "$BRANCH" =~ ^release-[0-9]+\.[0-9]+\.x$ ]]; then
-          echo "Detected release branch: $BRANCH"
-
-          # Check if matching branch exists in client repo
-          if ! git ls-remote --exit-code --heads https://github.com/ogx-ai/ogx-client-python.git "$BRANCH" > /dev/null 2>&1; then
-            echo "::error::Branch $BRANCH not found in ogx-client-python repository"
-            echo "::error::Please create the matching release branch in ogx-client-python before testing"
-            exit 1
-          fi
-
+          echo "Detected release branch: $BRANCH — generating client from checkout"
+          echo "generate=true" >> $GITHUB_OUTPUT
           echo "install-after-sync=true" >> $GITHUB_OUTPUT
-          echo "install-source=git+https://github.com/ogx-ai/ogx-client-python.git@$BRANCH" >> $GITHUB_OUTPUT
+          echo "install-source=${{ github.workspace }}/client-sdks/openapi/sdks/python" >> $GITHUB_OUTPUT
         elif [ "${{ inputs.client-version }}" = "latest" ]; then
-          # Install from main git after sync
+          echo "Generating client from checkout (client-version=latest)"
+          echo "generate=true" >> $GITHUB_OUTPUT
           echo "install-after-sync=true" >> $GITHUB_OUTPUT
-          echo "install-source=git+https://github.com/ogx-ai/ogx-client-python.git@main" >> $GITHUB_OUTPUT
+          echo "install-source=${{ github.workspace }}/client-sdks/openapi/sdks/python" >> $GITHUB_OUTPUT
         elif [ "${{ inputs.client-version }}" = "published" ]; then
           # Use published version from PyPI (installed by sync)
+          echo "generate=false" >> $GITHUB_OUTPUT
           echo "install-after-sync=false" >> $GITHUB_OUTPUT
         elif [ -n "${{ inputs.client-version }}" ]; then
           echo "::error::Invalid client-version: ${{ inputs.client-version }}"
           exit 1
+        else
+          echo "generate=false" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Set up Java (SDK generation)
+      if: steps.configure.outputs.generate == 'true'
+      uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654 # v5.2.0
+      with:
+        distribution: 'temurin'
+        java-version: '11'
+
+    - name: Set up Node.js (SDK generation)
+      if: steps.configure.outputs.generate == 'true'
+      uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
+      with:
+        node-version: '20'
+
+    - name: Install openapi-generator-cli (SDK generation)
+      if: steps.configure.outputs.generate == 'true'
+      shell: bash
+      run: npm install -g @openapitools/openapi-generator-cli
+
+    - name: Generate ogx-client SDK
+      if: steps.configure.outputs.generate == 'true'
+      shell: bash
+      working-directory: client-sdks/openapi
+      run: |
+        echo "Generating ogx-client SDK (OPEN=0) from checkout..."
+        make sdk OPEN=0
+
+        PY_FILE_COUNT=$(find sdks/python -name "*.py" | wc -l)
+        echo "Generated Python files: $PY_FILE_COUNT"
+        if [ "$PY_FILE_COUNT" -le 10 ]; then
+          echo "::error::Too few Python files generated (expected > 10, got $PY_FILE_COUNT)"
+          exit 1
         fi

.github/actions/launch-gpu-runner/action.yml

@@ -0,0 +1,82 @@
+name: 'Launch GPU EC2 Runner'
+description: 'Launch GPU-enabled EC2 instance as GitHub Actions self-hosted runner (wrapper for machulav/ec2-github-runner)'
+
+on:
+  workflow_dispatch:
+inputs:
+  mode:
+    description: 'Mode: start or stop'
+    required: true
+  github-token:
+    description: 'GitHub Personal Access Token with repo scope for runner registration'
+    required: true
+  instance-type:
+    description: 'EC2 instance type (e.g., g6.2xlarge, g5.2xlarge, g6.8xlarge)'
+    required: false
+    default: 'g6.2xlarge'
+  aws-region:
+    description: 'AWS region'
+    required: true
+  availability-zones-config:
+    description: 'JSON array of AZ configs with imageId, subnetId, securityGroupId for fallback'
+    required: false
+    default: ''
+  # For stop mode
+  label:
+    description: 'Runner label (for stop mode)'
+    required: false
+  ec2-instance-id:
+    description: 'EC2 instance ID (for stop mode)'
+    required: false
+  # Optional
+  ec2-instance-tags:
+    description: 'JSON array of tags to apply to EC2 instance'
+    required: false
+    default: '[]'
+  runner-home-dir:
+    description: 'Home directory for the runner'
+    required: false
+    default: ''
+  iam-role-name:
+    description: 'IAM role name to attach to the instance (optional, for enhanced security)'
+    required: false
+    default: ''
+
+outputs:
+  label:
+    description: 'Unique label for the launched runner'
+    value: ${{ steps.ec2-runner.outputs.label }}
+  ec2-instance-id:
+    description: 'EC2 instance ID'
+    value: ${{ steps.ec2-runner.outputs.ec2-instance-id }}
+
+runs:
+  using: 'composite'
+  steps:
+    - name: Start EC2 runner
+      if: inputs.mode == 'start'
+      id: ec2-runner
+      uses: machulav/ec2-github-runner@343a1b2ae682e681c3cec9a235d882da17ff04ef # v2.6.1
+      env:
+        AWS_DEFAULT_REGION: ${{ inputs.aws-region }}
+        AWS_REGION: ${{ inputs.aws-region }}
+      with:
+        mode: start
+        github-token: ${{ inputs.github-token }}
+        ec2-instance-type: ${{ inputs.instance-type }}
+        aws-resource-tags: ${{ inputs.ec2-instance-tags }}
+        runner-home-dir: ${{ inputs.runner-home-dir }}
+        iam-role-name: ${{ inputs.iam-role-name }}
+        availability-zones-config: ${{ inputs.availability-zones-config }}
+
+    - name: Stop EC2 runner
+      if: inputs.mode == 'stop'
+      uses: machulav/ec2-github-runner@343a1b2ae682e681c3cec9a235d882da17ff04ef # v2.6.1
+      env:
+        AWS_DEFAULT_REGION: ${{ inputs.aws-region }}
+        AWS_REGION: ${{ inputs.aws-region }}
+      with:
+        mode: stop
+        github-token: ${{ inputs.github-token }}
+        label: ${{ inputs.label }}
+        ec2-instance-id: ${{ inputs.ec2-instance-id }}

.github/actions/run-and-record-tests/action.yml

@@ -24,6 +24,10 @@ inputs:
     description: 'Regex pattern to pass to pytest -k'
     required: false
     default: ''
+  text-model:
+    description: 'Text model override to pass to integration-tests.sh'
+    required: false
+    default: ''
   target-branch:
     description: 'Target branch for recording commits (for PRs, use the PR head branch)'
     required: false
@@ -56,28 +60,37 @@ runs:
         INPUT_SUITE: ${{ inputs.suite }}
         INPUT_SUBDIRS: ${{ inputs.subdirs }}
         INPUT_PATTERN: ${{ inputs.pattern }}
+        INPUT_TEXT_MODEL: ${{ inputs.text-model }}
       run: |
-        SCRIPT_ARGS="--stack-config ${INPUT_STACK_CONFIG} --inference-mode ${INPUT_INFERENCE_MODE}"
+        SCRIPT_ARGS=(
+          --stack-config "${INPUT_STACK_CONFIG}"
+          --inference-mode "${INPUT_INFERENCE_MODE}"
+        )
 
         # Add optional arguments only if they are provided
         if [ -n "${INPUT_SETUP}" ]; then
-          SCRIPT_ARGS="${SCRIPT_ARGS} --setup ${INPUT_SETUP}"
+          SCRIPT_ARGS+=(--setup "${INPUT_SETUP}")
         fi
         if [ -n "${INPUT_SUITE}" ]; then
-          SCRIPT_ARGS="${SCRIPT_ARGS} --suite ${INPUT_SUITE}"
+          SCRIPT_ARGS+=(--suite "${INPUT_SUITE}")
         fi
         if [ -n "${INPUT_SUBDIRS}" ]; then
-          SCRIPT_ARGS="${SCRIPT_ARGS} --subdirs ${INPUT_SUBDIRS}"
+          SCRIPT_ARGS+=(--subdirs "${INPUT_SUBDIRS}")
         fi
         if [ -n "${INPUT_PATTERN}" ]; then
-          SCRIPT_ARGS="${SCRIPT_ARGS} --pattern ${INPUT_PATTERN}"
+          SCRIPT_ARGS+=(--pattern "${INPUT_PATTERN}")
+        fi
+        if [ -n "${INPUT_TEXT_MODEL}" ]; then
+          SCRIPT_ARGS+=(--text-model "${INPUT_TEXT_MODEL}")
         fi
 
         echo "=== Running command ==="
-        echo "uv run --no-sync ./scripts/integration-tests.sh $SCRIPT_ARGS"
+        printf 'uv run --no-sync ./scripts/integration-tests.sh'
+        printf ' %q' "${SCRIPT_ARGS[@]}"
+        printf '\n'
         echo ""
 
-        uv run --no-sync ./scripts/integration-tests.sh $SCRIPT_ARGS | tee "pytest-${INPUT_INFERENCE_MODE}.log"
+        uv run --no-sync ./scripts/integration-tests.sh "${SCRIPT_ARGS[@]}" | tee "pytest-${INPUT_INFERENCE_MODE}.log"
 
 
     - name: Commit and push recordings

.github/actions/setup-runner/action.yml

@@ -6,7 +6,7 @@ inputs:
     required: false
     default: "3.12"
   client-version:
-    description: The ogx-client-python version to test against (latest or published)
+    description: Which ogx-client to test against — 'latest' generates the SDK in-repo from client-sdks/openapi, 'published' uses the PyPI release
     required: false
     default: "latest"
   sdk_install_url:
@@ -46,7 +46,7 @@ runs:
         # Install specific client version after sync if needed
         if [ "${{ steps.client-config.outputs.install-after-sync }}" = "true" ]; then
           echo "Installing ogx-client from: ${{ steps.client-config.outputs.install-source }}"
-          uv pip install ${{ steps.client-config.outputs.install-source }}
+          uv pip install --upgrade ${{ steps.client-config.outputs.install-source }}
         fi
 
         echo "Installed ogx packages"

.github/actions/setup-test-environment/action.yml

@@ -27,6 +27,10 @@ inputs:
     description: 'Explicit branch override (used by scheduled CI to pass the matrix branch)'
     required: false
     default: ''
+  enable-hf-cache:
+    description: 'Whether to cache HuggingFace models and datasets'
+    required: false
+    default: 'true'
 
 runs:
   using: 'composite'
@@ -40,6 +44,7 @@ runs:
         branch: ${{ inputs.branch }}
 
     - name: Cache HuggingFace models and datasets
+      if: ${{ inputs.enable-hf-cache == 'true' }}
       uses: actions/cache@v4
       with:
         path: ~/.cache/huggingface
@@ -48,6 +53,7 @@ runs:
           hf-cache-${{ runner.os }}-
 
     - name: Pre-download HuggingFace assets for offline use
+      if: ${{ inputs.enable-hf-cache == 'true' }}
       shell: bash
       env:
         HF_HUB_ENABLE_HF_TRANSFER: "1"
@@ -74,7 +80,7 @@ runs:
         setup: ${{ inputs.setup }}
 
     - name: Setup vllm
-      if: ${{ startsWith(inputs.setup, 'vllm') && inputs.inference-mode != 'replay' }}
+      if: ${{ startsWith(inputs.setup, 'vllm') && !startsWith(inputs.setup, 'vllm-gpu-') && inputs.inference-mode != 'replay' }}
       uses: ./.github/actions/setup-vllm
 
     - name: Start Postgres service