feat: add greptimedb-quickstart skill with onboarding card

.gitignore

@@ -26,3 +26,7 @@ src/**/*.js
 # Exception: CodeBlock component is intentionally JavaScript
 !src/theme/CodeBlock/index.js
 variables/*.js
+
+# Generated by scripts/sync-skill.mjs from skills/*/SKILL.md
+static/SKILL.md
+static/skills/

docs/getting-started/overview.md

@@ -4,8 +4,12 @@ keywords: [getting started, overview, installation, quick start]
 description: Get started with GreptimeDB quickly.
 ---
 
+import AgentOnboarding from '@site/src/components/AgentOnboarding';
+
 # Getting Started
 
+<AgentOnboarding />
+
 Get started with GreptimeDB quickly by following these steps:
 
 - [Installation](./installation/overview.md): Learn how to install GreptimeDB as a standalone or cluster.

i18n/zh/docusaurus-plugin-content-docs/current/getting-started/overview.md

@@ -4,8 +4,12 @@ keywords: [快速开始, 安装]
 description: 快速开始使用 GreptimeDB
 ---
 
+import AgentOnboarding from '@site/src/components/AgentOnboarding';
+
 # 立即开始
 
+<AgentOnboarding />
+
 立即开始使用 GreptimeDB！
 
 - [安装](./installation/overview.md)：安装 GreptimeDB 单机模式或分布式集群。

i18n/zh/docusaurus-plugin-content-docs/version-1.0/getting-started/overview.md

@@ -4,8 +4,12 @@ keywords: [快速开始, 安装]
 description: 快速开始使用 GreptimeDB
 ---
 
+import AgentOnboarding from '@site/src/components/AgentOnboarding';
+
 # 立即开始
 
+<AgentOnboarding />
+
 立即开始使用 GreptimeDB！
 
 - [安装](./installation/overview.md)：安装 GreptimeDB 单机模式或分布式集群。

package.json

@@ -4,8 +4,11 @@
   "private": true,
   "scripts": {
     "docusaurus": "docusaurus",
+    "sync-skill": "node scripts/sync-skill.mjs",
+    "prestart": "pnpm sync-skill",
     "start": "docusaurus start",
     "start:zh": "DOC_LANG=zh pnpm start",
+    "prebuild": "pnpm sync-skill",
     "build": "docusaurus build",
     "swizzle": "docusaurus swizzle",
     "deploy": "docusaurus deploy",

scripts/sync-skill.mjs

@@ -0,0 +1,91 @@
+#!/usr/bin/env node
+// Mirrors every skills/<name>/SKILL.md to static/ so Docusaurus publishes them
+// at the site root:
+//   skills/greptimedb-quickstart/SKILL.md  -> static/SKILL.md
+//                                         -> static/skills/greptimedb-quickstart/SKILL.md
+//   skills/<other>/SKILL.md                 -> static/skills/<other>/SKILL.md
+//
+// The root /SKILL.md is the agent-onboarding entrypoint (copied from
+// greptimedb-quickstart). The /skills/<name>/SKILL.md endpoints let agents
+// fetch sister skills directly from the docs site instead of guessing GitHub
+// paths.
+//
+// Runs as a prestart / prebuild npm hook. Never edit anything under
+// static/SKILL.md or static/skills/ by hand.
+
+import { existsSync, mkdirSync, readdirSync, readFileSync, rmSync, writeFileSync } from 'node:fs';
+import { dirname, resolve } from 'node:path';
+import { fileURLToPath } from 'node:url';
+
+const repoRoot = resolve(dirname(fileURLToPath(import.meta.url)), '..');
+const skillsDir = resolve(repoRoot, 'skills');
+const staticDir = resolve(repoRoot, 'static');
+const ROOT_SKILL = 'greptimedb-quickstart';
+
+// Wipe previous outputs so a renamed or deleted skill does not leave a stale
+// SKILL.md behind under static/skills/.
+const staticSkillsDir = resolve(staticDir, 'skills');
+const staticRootSkill = resolve(staticDir, 'SKILL.md');
+if (existsSync(staticSkillsDir)) rmSync(staticSkillsDir, { recursive: true, force: true });
+if (existsSync(staticRootSkill)) rmSync(staticRootSkill, { force: true });
+
+// Load version variables from variables/variables-<latestStable>.ts and resolve
+// VAR::name placeholders in each skill, mirroring what llms-txt-generator.ts
+// does for the .md endpoints. Without this, version-pinned commands in
+// SKILL.md (Docker tag, install.sh argument) would ship the literal
+// "VAR::greptimedbVersion" string to agents.
+const latestVersion = JSON.parse(readFileSync(resolve(repoRoot, 'versions.json'), 'utf8'))[0];
+const variables = loadVariables(latestVersion);
+
+function loadVariables(version) {
+  const file = resolve(repoRoot, 'variables', `variables-${version}.ts`);
+  if (!existsSync(file)) {
+    console.warn(`[sync-skill] variables-${version}.ts not found; VAR:: placeholders will not be resolved`);
+    return {};
+  }
+  const text = readFileSync(file, 'utf8');
+  const out = {};
+  for (const m of text.matchAll(/(\w+):\s*['"]([^'"]+)['"]/g)) {
+    out[m[1]] = m[2];
+  }
+  return out;
+}
+
+function resolveVariables(content) {
+  return content.replace(/VAR::([A-Z_]+)/gi, (match, name) => variables[name] ?? match);
+}
+
+// Inject the breadcrumb as a YAML comment inside the frontmatter so strict
+// frontmatter parsers (Anthropic Skill loaders, etc.) still see `---` as the
+// first byte of the file. Falls back to an HTML comment prefix when the
+// source has no frontmatter.
+function withBanner(body, srcRel) {
+  const yamlComment = `# Generated from ${srcRel}. Do not edit by hand.\n`;
+  const match = body.match(/^---\r?\n/);
+  if (match) {
+    return body.slice(0, match[0].length) + yamlComment + body.slice(match[0].length);
+  }
+  return `<!-- Generated from ${srcRel}. Do not edit by hand. -->\n` + body;
+}
+
+function writeWithBanner(srcAbs, dstAbs, srcRel) {
+  mkdirSync(dirname(dstAbs), { recursive: true });
+  const body = resolveVariables(readFileSync(srcAbs, 'utf8'));
+  writeFileSync(dstAbs, withBanner(body, srcRel), 'utf8');
+  console.log(`[sync-skill] ${srcRel} -> ${dstAbs.slice(repoRoot.length + 1)}`);
+}
+
+for (const name of readdirSync(skillsDir, { withFileTypes: true })) {
+  if (!name.isDirectory()) continue;
+  const src = resolve(skillsDir, name.name, 'SKILL.md');
+  const srcRel = `skills/${name.name}/SKILL.md`;
+  try {
+    readFileSync(src);
+  } catch {
+    continue;
+  }
+  writeWithBanner(src, resolve(staticDir, 'skills', name.name, 'SKILL.md'), srcRel);
+  if (name.name === ROOT_SKILL) {
+    writeWithBanner(src, resolve(staticDir, 'SKILL.md'), srcRel);
+  }
+}

skills/README.md

@@ -6,14 +6,23 @@ This is the GreptimeDB documentation for AI agents.
 
 ## Content
 
+- `greptimedb-quickstart`: Entry-point guide — when to use GreptimeDB, how to install, which write protocol to choose, how to query, plus pointers to deeper docs via `llms.txt`. Start here.
 - `greptimedb-pipeline`: For creating greptimedb pipeline definition
 - `greptimedb-flow`: For creating greptimedb flow, continuous aggregation tasks
 - `greptimedb-trigger`: For creating greptimedb trigger
 
+The `greptimedb-quickstart` skill is also hosted at <https://docs.greptime.com/SKILL.md> (and <https://docs.greptime.cn/SKILL.md>), so any AI coding agent can load it with a single instruction:
+
+> Read https://docs.greptime.com/SKILL.md and follow the instructions to use GreptimeDB with your AI agent — deploy, configure, ingest, and query.
+
 ## How to install
 
 Using `skills` cli tool to install the skill to your coding agents.
 
+### `greptimedb-quickstart`
+
+`npx skills add https://github.com/GreptimeTeam/docs/tree/main/skills/greptimedb-quickstart`
+
 ### `greptimedb-pipeline`
 
 `npx skills add https://github.com/GreptimeTeam/docs/tree/main/skills/greptimedb-pipeline`