### Local Development Setup

Source: https://github.com/llm-d/llm-d.github.io/blob/main/CONTRIBUTING.md

Steps to set up the project locally for development, including cloning, installing dependencies, and starting the development server.

```bash
git clone https://github.com/YOUR-USERNAME/llm-d.github.io.git
cd llm-d.github.io
npm install
```

```bash
npm start
```

```bash
git add .
git commit -s -m "docs: your change description"
git push origin docs/your-change-description
```

--------------------------------

### Install and Start Local Development Server

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/README.md

Installs project dependencies and starts a local development server for previewing documentation changes.

```bash
npm install
npm start
```

--------------------------------

### Start Local Development Server

Source: https://github.com/llm-d/llm-d.github.io/blob/main/CONTRIBUTING.md

Run this command to start the local development server. It opens a browser with live reload, reflecting most changes immediately.

```bash
npm start
```

--------------------------------

### Install Project Dependencies

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Run this command to install all necessary Node.js dependencies for the project.

```bash
npm install
```

--------------------------------

### Run Link Checker Commands

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Build the site and then run the link checker. The checker starts a local server, crawls pages, checks links, generates a report, and stops the server.

```bash
# 1. Build the site first
npm run build:all

# 2. Run the link checker
npm run check-links
```

--------------------------------

### Custom Tabs with Bash Example

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/scripts/test-fixtures/transformation-test.md

Demonstrates the transformation of custom tabs, including a default tab with a bash code block.

```bash
echo "First tab code block"
```

--------------------------------

### Nested Content in Tabs: Google GKE Example

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/scripts/test-fixtures/transformation-test.md

Shows a bash command within the Google GKE tab, demonstrating nested content transformation.

```bash
gcloud container clusters create my-cluster
```

--------------------------------

### Single Tab Example

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/scripts/test-fixtures/transformation-test.expected.md

Illustrates a Tabs component with a single tab. This tests the rendering of a minimal tab configuration.

```jsx
<Tabs>

<TabItem value="only-tab" label="Only Tab" default>
This is the only tab.

</TabItem>
</Tabs>

```

--------------------------------

### Empty Tab Example

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/scripts/test-fixtures/transformation-test.expected.md

Demonstrates an empty tab within a Tabs component. This tests how the system handles tabs with no content.

```jsx
<Tabs>

<TabItem value="empty-tab" label="Empty Tab" default>

</TabItem>
</Tabs>

```

--------------------------------

### Install llm-d Skills for AI Coding Agents

Source: https://github.com/llm-d/llm-d.github.io/blob/main/blog/2026-06-23_networking-for-distributed-inference-llm-d.md

Clone the llm-d-pd-utils repository and copy the skills into your AI agent's skill directory to enable functionalities like running preflight checks and networking tests.

```bash
AGENT_SKILLS_DIR=~/.claude/skills/
git clone https://github.com/llm-d/llm-d-pd-utils.git /tmp/llmd-skills-repo
cp -r /tmp/llmd-skills-repo/skills/* $AGENT_SKILLS_DIR
rm -rf /tmp/llmd-skills-repo
```

--------------------------------

### Get Session Start Epoch Time

Source: https://github.com/llm-d/llm-d.github.io/blob/main/community/events.md

This JavaScript function parses date and time strings from session data to calculate the session's start time as an epoch timestamp. It handles various date and time formats and returns a large integer if parsing fails.

```javascript
const getSessionStartEpoch = (s) => {
    try {
      const monthMap = {
        Jan: 0, Feb: 1, Mar: 2, Apr: 3, May: 4, Jun: 5,
        Jul: 6, Aug: 7, Sep: 8, Oct: 9, Nov: 10, Dec: 11,
      };
      const dateText = String(s.date || '');
      const timeText = String(s.time || '');

      const dateMatch = dateText.match(/([A-Za-z]{3})\\\s+(\\\d{1,2}),\\s*(\\\d{4})/);
      if (!dateMatch) return Number.MAX_SAFE_INTEGER;
      const monthIdx = monthMap[dateMatch[1]];
      const day = parseInt(dateMatch[2], 10);
      const year = parseInt(dateMatch[3], 10);

      const timeMatch = timeText.match(/(\\\d{1,2})(?::(\\\d{2}))?\\\s*(am|pm)/i);
      const hour12 = timeMatch ? parseInt(timeMatch[1] || '0', 10) : 0;
      const minute = timeMatch && timeMatch[2] ? parseInt(timeMatch[2], 10) : 0;
      const meridiem = timeMatch ? (timeMatch[3] || '').toLowerCase() : 'am';
      let hour24 = hour12 % 12;
      if (meridiem === 'pm') hour24 += 12;

```

--------------------------------

### Local Testing Workflow: Build and Serve

Source: https://github.com/llm-d/llm-d.github.io/blob/main/scripts/README.md

This command builds the entire site and serves it locally for testing. Open your browser to http://localhost:3000 for the main site and http://localhost:3000/docs for the docs site.

```bash
npm run serve:production
```

--------------------------------

### Build Documentation with Local Clone and Fetch Latest

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/README.md

Builds documentation using a local llm-d clone, pulling the latest changes from the origin before building.

```bash
# Use local clone but pull latest from origin first
LLMD_REPO=~/repos/llm-d LLMD_FETCH=1 npm run build:all
```

--------------------------------

### Link Checker Report Example

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

An example of the broken links report format. It lists broken links grouped by source page, indicating the type of issue and the repository to fix.

```markdown
### /videos

**Source:** Local (this repository)

- 🔗 `/docs/guide` → **HTTP 404** (link)
```

--------------------------------

### Sync Documentation and Build

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Syncs documentation from the llm-d repository and then builds the project. This is an alternative to `build:all` when you want to sync separately before building.

```bash
# Or sync only, then build
cd preview
LLMD_REPO=~/repos/llm-d bash scripts/sync-docs.sh
npm run build
```

--------------------------------

### Build All Content for Full Site Preview

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Builds the entire site, including all synced documentation from the llm-d GitHub repository. This command clones the repository if LLMD_REPO is not set.

```bash
npm run build:all
```

--------------------------------

### Perform Full Project Build

Source: https://github.com/llm-d/llm-d.github.io/blob/main/CONTRIBUTING.md

Execute this command to generate static content into the `build` directory. This command tests the complete build process, including the main site, preview docs sync, and merging.

```bash
npm run build
```

--------------------------------

### Build All Project Assets

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Builds all project assets, including documentation, using a local clone of the llm-d repository. This is recommended for testing sync scripts without network access.

```bash
# Using a local llm-d clone (recommended — no network required)
LLMD_REPO=~/repos/llm-d npm run build:all
```

--------------------------------

### Send SLO-Aware Request to Inference Gateway

Source: https://github.com/llm-d/llm-d.github.io/blob/main/blog/2026-03-13_predicted-latency-based-scheduling-for-llms.md

Send a completion request to the Inference Gateway with custom headers for prediction-based scheduling and SLOs. This example demonstrates how to set time-to-first-byte (TTFT) and time-to-process-all-output (TPOT) targets.

```bash
curl -v $GW_IP/v1/completions \
  -H 'Content-Type: application/json' \
  -H 'x-prediction-based-scheduling: true' \
  -H 'x-slo-ttft-ms: 200' \
  -H 'x-slo-tpot-ms: 50' \
  -d '{
    "model": "Qwen/Qwen3-32B",
    "prompt": "what is the difference between Franz and Apache Kafka?",
    "max_tokens": 200,
    "temperature": 0,
    "stream_options": {"include_usage": "true"},
    "stream": "true"
  }'
```

--------------------------------

### Build Documentation with Local llm-d Clone

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/README.md

Builds documentation using a local clone of the llm-d repository, which is faster and does not require a network connection.

```bash
# Use a local llm-d clone (fast, no network required)
LLMD_REPO=~/repos/llm-d npm run build:all
```

--------------------------------

### Sync Documentation Only

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/README.md

Syncs documentation from the upstream llm-d repository without performing a full build. Run this command from the preview/ directory.

```bash
# From the preview/ directory
LLMD_REPO=~/repos/llm-d bash scripts/sync-docs.sh
```

--------------------------------

### Nested Content in Tabs: AWS EKS Example

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/scripts/test-fixtures/transformation-test.md

Illustrates nested markdown elements within tabs, specifically a YAML code block, link, and note callout within the AWS EKS tab.

```yaml
apiVersion: v1
kind: Service
metadata:
  name: test-service
```

--------------------------------

### Configure Container Startup with Preflight Checks

Source: https://github.com/llm-d/llm-d.github.io/blob/main/blog/2026-06-23_networking-for-distributed-inference-llm-d.md

Modify the container's startup arguments to include a preflight check script that runs before vLLM. The `&&` operator ensures that `vllm serve` only executes if the preflight checks pass.

```yaml
containers:
  - args:
      - "... python3 llm-d-preflight-checks.py && vllm serve /model-cache/models/…"
```

--------------------------------

### Set Environment Variable to Pause Execution for Complex Tests

Source: https://github.com/llm-d/llm-d.github.io/blob/main/blog/2026-06-23_networking-for-distributed-inference-llm-d.md

Use the `LLMD_PREFLIGHT_CHECKS` environment variable set to `pause` to halt execution before vLLM starts, allowing for more complex networking tests with full GPU memory availability.

```yaml
env:
  - name: LLMD_PREFLIGHT_CHECKS
    value: pause
```

--------------------------------

### Build All Content Using Local llm-d Clone with Fetch

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Builds the entire site using a local clone of the llm-d repository and fetches the latest changes from origin first. Set LLMD_REPO to the local path and LLMD_FETCH to 1.

```bash
LLMD_REPO=~/repos/llm-d LLMD_FETCH=1 npm run build:all
```

--------------------------------

### Create Version Snapshot Script

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/VERSIONING.md

Execute this script in the 'preview' directory to create a new version snapshot. This syncs documentation from the release branch and updates versioning files.

```bash
cd preview
./scripts/create-version.sh 0.7
```

--------------------------------

### Build All Documentation from Upstream Repo

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/README.md

Builds all documentation, syncing from the upstream llm-d repository. This is suitable for CI/production environments.

```bash
# Clone from GitHub and build everything (CI/production)
npm run build:all
```

--------------------------------

### Commit and Deploy Documentation Changes

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/VERSIONING.md

After updating the configuration and creating the version snapshot, stage, commit, and push the changes to deploy the new documentation version.

```bash
git add versioned_docs/ versioned_sidebars/ versions.json preview/docusaurus.config.ts
git commit -m "Add v0.7 documentation version"
git push
```

--------------------------------

### Build All Content Using Local llm-d Clone

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Builds the entire site using a local clone of the llm-d repository, skipping network operations. Set the LLMD_REPO environment variable to the path of your local clone.

```bash
LLMD_REPO=~/repos/llm-d npm run build:all
```

--------------------------------

### Troubleshooting: Sync Docs Manually

Source: https://github.com/llm-d/llm-d.github.io/blob/main/scripts/README.md

To resolve issues with stale docs content or build failures in the preview step, you can manually sync the docs. Navigate to the preview directory and run the sync script.

```bash
cd preview && bash scripts/sync-docs.sh main
```

--------------------------------

### Serve the Built Site

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Serves the static site that has been generated by the build process. This is typically used after running a build command.

```bash
npm run serve
```

--------------------------------

### Run Create Release Branch Workflow

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/RELEASE-BRANCH-SYNC.md

Manually trigger the creation of a new release branch for documentation. Specify the full version number and the source branch from the main llm-d repository.

```bash
gh workflow run create-release-branch.yml \
  --repo llm-d/llm-d.github.io \
  --field version=0.7.0 \
  --field source_branch=release-0.7
```

--------------------------------

### Create Additional Version Snapshots

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/VERSIONING.md

To create snapshots for older versions (e.g., 0.6, 0.5), ensure the corresponding release branches exist in the main repository and then run the 'create-version.sh' script with the desired version number.

```bash
# Requires release-0.6, release-0.5 branches in llm-d/llm-d
./scripts/create-version.sh 0.6
./scripts/create-version.sh 0.5
```

--------------------------------

### Create Remote Source Configuration

Source: https://github.com/llm-d/llm-d.github.io/blob/main/CONTRIBUTING.md

Define a new JavaScript configuration file for community content, specifying source URL, output directory, and content transformation.

```javascript
import { createContentWithSource, createStandardTransform, getLlmdRepoConfig } from '../utils.js';

const { sourceBaseUrl } = getLlmdRepoConfig();
const contentTransform = createStandardTransform();

export default [
  'docusaurus-plugin-remote-content',
  {
    name: 'governance',
    sourceBaseUrl,
    outDir: 'community',
    documents: ['GOVERNANCE.md'],
    noRuntimeDownloads: false,
    performCleanup: true,

    modifyContent(filename, content) {
      if (filename === 'GOVERNANCE.md') {
        return createContentWithSource({
          title: 'Project Governance',
          description: 'Governance structure for the llm-d project',
          sidebarLabel: 'Governance',
          sidebarPosition: 6,
          filename: 'GOVERNANCE.md',
          newFilename: 'governance.md',
          content,
          contentTransform
        });
      }
      return undefined;
    },
  },
];
```

--------------------------------

### Run Transformation Tests

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/scripts/test-fixtures/README.md

Execute the transformation tests from the root or preview directory. These commands are used to verify the markdown transformation scripts.

```bash
# From root directory
npm run test:transformations

# Or from preview directory
cd preview && npm run test:transformations

# Run all tests (Jest + transformations)
npm test
```

--------------------------------

### List Sync Release Docs Workflow Runs

Source: https://github.com/llm-d/llm-d.github.io/blob/main/preview/RELEASE-BRANCH-SYNC.md

Check the logs for the sync-release-docs.yml workflow to troubleshoot sync failures. This command lists recent runs of the specified workflow for the llm-d/llm-d.github.io repository.

```bash
gh run list --workflow=sync-release-docs.yml --repo llm-d/llm-d.github.io
```

--------------------------------

### JavaScript Configuration for Contributing File

Source: https://github.com/llm-d/llm-d.github.io/blob/main/README.md

Configuration file for syncing the CONTRIBUTING.md file from the main repository to the website's community documentation section.

```javascript
remote-content/remote-sources/community/contribute.js
```