@better-openclaw/core 1.0.0

package/src/generators/skills.ts

@@ -0,0 +1,628 @@
+import Handlebars from "handlebars";
+import type { ResolverOutput } from "../types.js";
+
+/**
+ * Basic skill templates keyed by skillId.
+ *
+ * Templates use Handlebars `{{VAR}}` syntax. Variables are resolved from the
+ * service's `openclawEnvVars` array at generation time.
+ */
+const skillTemplates: Record<string, string> = {
+	"redis-cache": `---
+name: redis-cache
+description: "Cache data and manage key-value state using Redis"
+metadata:
+  openclaw:
+    emoji: "🔴"
+---
+
+# Redis Cache
+
+Use Redis as a high-performance in-memory cache and key-value store.
+
+## Connection Details
+
+- **Host:** \`{{REDIS_HOST}}\`
+- **Port:** \`{{REDIS_PORT}}\`
+
+## Example Commands
+
+### Set a value
+\`\`\`bash
+redis-cli -h {{REDIS_HOST}} -p {{REDIS_PORT}} -a $REDIS_PASSWORD SET mykey "myvalue"
+\`\`\`
+
+### Get a value
+\`\`\`bash
+redis-cli -h {{REDIS_HOST}} -p {{REDIS_PORT}} -a $REDIS_PASSWORD GET mykey
+\`\`\`
+
+### List all keys
+\`\`\`bash
+redis-cli -h {{REDIS_HOST}} -p {{REDIS_PORT}} -a $REDIS_PASSWORD KEYS "*"
+\`\`\`
+
+## Usage Notes
+
+- Use Redis for caching frequently accessed data, session storage, and pub/sub messaging.
+- Data is persisted to disk via RDB snapshots in the mounted volume.
+- Password authentication is required.
+`,
+
+	"qdrant-memory": `---
+name: qdrant-memory
+description: "Store and retrieve vector embeddings for semantic search and RAG"
+metadata:
+  openclaw:
+    emoji: "🧠"
+---
+
+# Qdrant Memory
+
+Use Qdrant as a vector database for storing embeddings, enabling semantic search and retrieval-augmented generation (RAG).
+
+## Connection Details
+
+- **Host:** \`{{QDRANT_HOST}}\`
+- **Port:** \`{{QDRANT_PORT}}\` (REST API)
+
+## Example API Calls
+
+### Create a collection
+\`\`\`bash
+curl -X PUT "http://{{QDRANT_HOST}}:{{QDRANT_PORT}}/collections/my_collection" \\
+  -H "Content-Type: application/json" \\
+  -d '{"vectors": {"size": 384, "distance": "Cosine"}}'
+\`\`\`
+
+### Upsert points
+\`\`\`bash
+curl -X PUT "http://{{QDRANT_HOST}}:{{QDRANT_PORT}}/collections/my_collection/points" \\
+  -H "Content-Type: application/json" \\
+  -d '{"points": [{"id": 1, "vector": [0.1, 0.2, ...], "payload": {"text": "hello"}}]}'
+\`\`\`
+
+### Search similar vectors
+\`\`\`bash
+curl -X POST "http://{{QDRANT_HOST}}:{{QDRANT_PORT}}/collections/my_collection/points/search" \\
+  -H "Content-Type: application/json" \\
+  -d '{"vector": [0.1, 0.2, ...], "limit": 5}'
+\`\`\`
+
+## Usage Notes
+
+- Use 384-dimensional vectors for MiniLM-based embeddings or 1536 for OpenAI ada-002.
+- Qdrant supports filtering, batch operations, and named vectors.
+`,
+
+	"n8n-trigger": `---
+name: n8n-trigger
+description: "Trigger and manage automation workflows using n8n"
+metadata:
+  openclaw:
+    emoji: "🔄"
+---
+
+# n8n Workflow Trigger
+
+Use n8n to create, trigger, and manage automation workflows via its REST API.
+
+## Connection Details
+
+- **Host:** \`{{N8N_HOST}}\`
+- **Port:** \`{{N8N_PORT}}\`
+- **Webhook URL:** \`{{N8N_WEBHOOK_URL}}\`
+
+## Example API Calls
+
+### Trigger a webhook workflow
+\`\`\`bash
+curl -X POST "{{N8N_WEBHOOK_URL}}webhook/<your-webhook-id>" \\
+  -H "Content-Type: application/json" \\
+  -d '{"message": "Hello from OpenClaw"}'
+\`\`\`
+
+### List workflows
+\`\`\`bash
+curl -X GET "http://{{N8N_HOST}}:{{N8N_PORT}}/api/v1/workflows" \\
+  -H "Content-Type: application/json"
+\`\`\`
+
+### Activate a workflow
+\`\`\`bash
+curl -X PATCH "http://{{N8N_HOST}}:{{N8N_PORT}}/api/v1/workflows/<id>" \\
+  -H "Content-Type: application/json" \\
+  -d '{"active": true}'
+\`\`\`
+
+## Usage Notes
+
+- Create workflows in the n8n UI at http://{{N8N_HOST}}:{{N8N_PORT}}.
+- Webhook nodes allow external triggers from OpenClaw.
+- n8n stores workflow state in PostgreSQL.
+`,
+
+	"ffmpeg-process": `---
+name: ffmpeg-process
+description: "Process media files using FFmpeg for transcoding, conversion, and manipulation"
+metadata:
+  openclaw:
+    emoji: "🎬"
+---
+
+# FFmpeg Media Processing
+
+Use FFmpeg for video/audio transcoding, format conversion, and media manipulation.
+
+## Shared Directory
+
+- **Media directory:** \`{{FFMPEG_SHARED_DIR}}\`
+
+## Example Commands
+
+### Convert video to MP4
+\`\`\`bash
+docker exec ffmpeg ffmpeg -i /data/input.avi -c:v libx264 -c:a aac /data/output.mp4
+\`\`\`
+
+### Extract audio from video
+\`\`\`bash
+docker exec ffmpeg ffmpeg -i /data/video.mp4 -vn -acodec libmp3lame /data/audio.mp3
+\`\`\`
+
+### Create thumbnail from video
+\`\`\`bash
+docker exec ffmpeg ffmpeg -i /data/video.mp4 -ss 00:00:05 -vframes 1 /data/thumb.jpg
+\`\`\`
+
+### Resize video
+\`\`\`bash
+docker exec ffmpeg ffmpeg -i /data/input.mp4 -vf scale=1280:720 /data/output_720p.mp4
+\`\`\`
+
+## Usage Notes
+
+- Place input files in the shared media directory.
+- FFmpeg runs as a sidecar container sharing a volume with OpenClaw.
+- Output files appear in the same shared directory.
+`,
+
+	"minio-storage": `---
+name: minio-storage
+description: "Store and retrieve files using S3-compatible object storage"
+metadata:
+  openclaw:
+    emoji: "💾"
+---
+
+# MinIO Object Storage
+
+Use MinIO as S3-compatible object storage for files, assets, and backups.
+
+## Connection Details
+
+- **Host:** \`{{MINIO_HOST}}\`
+- **API Port:** \`{{MINIO_PORT}}\`
+- **Access Key:** Uses \`MINIO_ACCESS_KEY\` env var
+- **Secret Key:** Uses \`MINIO_SECRET_KEY\` env var
+
+## Example API Calls
+
+### Create a bucket
+\`\`\`bash
+mc alias set local http://{{MINIO_HOST}}:{{MINIO_PORT}} $MINIO_ACCESS_KEY $MINIO_SECRET_KEY
+mc mb local/my-bucket
+\`\`\`
+
+### Upload a file
+\`\`\`bash
+mc cp /path/to/file.txt local/my-bucket/file.txt
+\`\`\`
+
+### Download a file
+\`\`\`bash
+mc cp local/my-bucket/file.txt /path/to/local/file.txt
+\`\`\`
+
+### List bucket contents
+\`\`\`bash
+mc ls local/my-bucket/
+\`\`\`
+
+## Usage Notes
+
+- MinIO is fully S3-compatible—use any S3 SDK or CLI.
+- Access the web console at http://{{MINIO_HOST}}:9001 for a visual file browser.
+- Create separate buckets for different data types (assets, backups, uploads).
+`,
+
+	"browserless-browse": `---
+name: browserless-browse
+description: "Automate browser interactions, scrape web pages, and generate PDFs"
+metadata:
+  openclaw:
+    emoji: "🌐"
+---
+
+# Browserless Browser Automation
+
+Use Browserless for headless Chrome browser automation, web scraping, screenshots, and PDF generation.
+
+## Connection Details
+
+- **Host:** \`{{BROWSERLESS_HOST}}\`
+- **Port:** \`{{BROWSERLESS_PORT}}\`
+- **Token:** Uses \`BROWSERLESS_TOKEN\` env var
+
+## Example API Calls
+
+### Take a screenshot
+\`\`\`bash
+curl -X POST "http://{{BROWSERLESS_HOST}}:{{BROWSERLESS_PORT}}/screenshot?token=$BROWSERLESS_TOKEN" \\
+  -H "Content-Type: application/json" \\
+  -d '{"url": "https://example.com", "options": {"fullPage": true}}' \\
+  --output screenshot.png
+\`\`\`
+
+### Generate a PDF
+\`\`\`bash
+curl -X POST "http://{{BROWSERLESS_HOST}}:{{BROWSERLESS_PORT}}/pdf?token=$BROWSERLESS_TOKEN" \\
+  -H "Content-Type: application/json" \\
+  -d '{"url": "https://example.com"}' \\
+  --output page.pdf
+\`\`\`
+
+### Scrape page content
+\`\`\`bash
+curl -X POST "http://{{BROWSERLESS_HOST}}:{{BROWSERLESS_PORT}}/content?token=$BROWSERLESS_TOKEN" \\
+  -H "Content-Type: application/json" \\
+  -d '{"url": "https://example.com"}'
+\`\`\`
+
+## Usage Notes
+
+- Browserless manages a pool of Chrome instances (max concurrent sessions configured via env).
+- Token authentication is required for all API calls.
+- Supports Puppeteer and Playwright WebSocket connections.
+`,
+
+	"searxng-search": `---
+name: searxng-search
+description: "Search the web using a privacy-respecting metasearch engine"
+metadata:
+  openclaw:
+    emoji: "🔍"
+---
+
+# SearXNG Web Search
+
+Use SearXNG as a privacy-respecting metasearch engine to query the web.
+
+## Connection Details
+
+- **Host:** \`{{SEARXNG_HOST}}\`
+- **Port:** \`{{SEARXNG_PORT}}\`
+
+## Example API Calls
+
+### Search the web
+\`\`\`bash
+curl "http://{{SEARXNG_HOST}}:{{SEARXNG_PORT}}/search?q=your+search+query&format=json"
+\`\`\`
+
+### Search with category filter
+\`\`\`bash
+curl "http://{{SEARXNG_HOST}}:{{SEARXNG_PORT}}/search?q=openai&categories=it&format=json"
+\`\`\`
+
+### Search with language filter
+\`\`\`bash
+curl "http://{{SEARXNG_HOST}}:{{SEARXNG_PORT}}/search?q=hello&language=en&format=json"
+\`\`\`
+
+## Usage Notes
+
+- Always use \`format=json\` for machine-readable results.
+- Available categories: general, images, videos, news, map, music, it, science, files.
+- SearXNG aggregates results from many search engines without tracking.
+`,
+
+	"whisper-transcribe": `---
+name: whisper-transcribe
+description: "Transcribe audio files to text using Faster Whisper"
+metadata:
+  openclaw:
+    emoji: "🎙️"
+---
+
+# Whisper Transcription
+
+Use the Faster Whisper server for speech-to-text transcription.
+
+## Connection Details
+
+- **Host:** \`{{WHISPER_HOST}}\`
+- **Port:** \`{{WHISPER_PORT}}\`
+
+## Example API Calls
+
+### Transcribe an audio file
+\`\`\`bash
+curl -X POST "http://{{WHISPER_HOST}}:{{WHISPER_PORT}}/v1/audio/transcriptions" \\
+  -F "file=@/path/to/audio.mp3" \\
+  -F "model=base"
+\`\`\`
+
+### Transcribe with language hint
+\`\`\`bash
+curl -X POST "http://{{WHISPER_HOST}}:{{WHISPER_PORT}}/v1/audio/transcriptions" \\
+  -F "file=@/path/to/audio.wav" \\
+  -F "model=base" \\
+  -F "language=en"
+\`\`\`
+
+### Get available models
+\`\`\`bash
+curl "http://{{WHISPER_HOST}}:{{WHISPER_PORT}}/v1/models"
+\`\`\`
+
+## Usage Notes
+
+- Supports MP3, WAV, FLAC, and other common audio formats.
+- Model sizes: tiny, base, small, medium, large (larger = more accurate but slower).
+- GPU acceleration significantly improves transcription speed.
+`,
+
+	"ollama-local-llm": `---
+name: ollama-local-llm
+description: "Run local language models for text generation, chat, and embeddings"
+metadata:
+  openclaw:
+    emoji: "🦙"
+---
+
+# Ollama Local LLM
+
+Use Ollama to run large language models locally for text generation, chat, and embeddings.
+
+## Connection Details
+
+- **Host:** \`{{OLLAMA_HOST}}\`
+- **Port:** \`{{OLLAMA_PORT}}\`
+
+## Example API Calls
+
+### Generate text
+\`\`\`bash
+curl -X POST "http://{{OLLAMA_HOST}}:{{OLLAMA_PORT}}/api/generate" \\
+  -H "Content-Type: application/json" \\
+  -d '{"model": "llama3.2", "prompt": "Explain quantum computing in simple terms"}'
+\`\`\`
+
+### Chat completion
+\`\`\`bash
+curl -X POST "http://{{OLLAMA_HOST}}:{{OLLAMA_PORT}}/api/chat" \\
+  -H "Content-Type: application/json" \\
+  -d '{"model": "llama3.2", "messages": [{"role": "user", "content": "Hello!"}]}'
+\`\`\`
+
+### Pull a model
+\`\`\`bash
+curl -X POST "http://{{OLLAMA_HOST}}:{{OLLAMA_PORT}}/api/pull" \\
+  -H "Content-Type: application/json" \\
+  -d '{"name": "llama3.2"}'
+\`\`\`
+
+### Generate embeddings
+\`\`\`bash
+curl -X POST "http://{{OLLAMA_HOST}}:{{OLLAMA_PORT}}/api/embeddings" \\
+  -H "Content-Type: application/json" \\
+  -d '{"model": "nomic-embed-text", "prompt": "Hello world"}'
+\`\`\`
+
+## Usage Notes
+
+- Pull models before first use (they are cached in the persistent volume).
+- Recommended models: llama3.2 (general), codellama (code), nomic-embed-text (embeddings).
+- GPU passthrough dramatically improves inference speed.
+- The Ollama API is OpenAI-compatible at /v1/ endpoints.
+`,
+
+	"remotion-render": `---
+name: remotion-render
+description: "Create and render videos programmatically using React"
+metadata:
+  openclaw:
+    emoji: "🎥"
+---
+
+# Remotion Video Rendering
+
+Use Remotion Studio to create and render videos programmatically with React.
+
+## Connection Details
+
+- **Host:** \`{{REMOTION_HOST}}\`
+- **Port:** \`{{REMOTION_PORT}}\`
+
+## Example Usage
+
+### Access the Studio UI
+Open \`http://{{REMOTION_HOST}}:{{REMOTION_PORT}}\` in your browser to use the Remotion Studio visual editor.
+
+### Render a video via CLI
+\`\`\`bash
+docker exec remotion npx remotion render src/index.tsx MyComposition out/video.mp4
+\`\`\`
+
+### Render with custom props
+\`\`\`bash
+docker exec remotion npx remotion render src/index.tsx MyComposition out/video.mp4 \\
+  --props='{"title": "Hello World"}'
+\`\`\`
+
+## Usage Notes
+
+- Define video compositions in React components.
+- Remotion supports MP4, WebM, and GIF output formats.
+- Use the Studio UI for previewing before rendering.
+- Combine with FFmpeg for post-processing.
+`,
+
+	"lightpanda-browse": `---
+name: lightpanda-browse
+description: "Browse the web using the ultra-fast LightPanda headless browser via CDP"
+metadata:
+  openclaw:
+    emoji: "🐼"
+---
+
+# LightPanda Browse
+
+LightPanda is an ultra-fast headless browser available via CDP WebSocket at \`ws://{{LIGHTPANDA_HOST}}:{{LIGHTPANDA_PORT}}\`.
+
+## Connect via Puppeteer
+
+\`\`\`javascript
+const browser = await puppeteer.connect({
+  browserWSEndpoint: "ws://{{LIGHTPANDA_HOST}}:{{LIGHTPANDA_PORT}}"
+});
+const page = await browser.newPage();
+await page.goto('https://example.com');
+const content = await page.evaluate(() => document.body.innerText);
+\`\`\`
+
+## Key Advantages
+
+- 9x less memory than Chrome (ideal for parallel scraping)
+- 11x faster page loading
+- Instant startup
+- Full CDP compatibility with Puppeteer and Playwright
+`,
+
+	"steel-browse": `---
+name: steel-browse
+description: "Browse the web using Steel Browser API with session management and anti-detection"
+metadata:
+  openclaw:
+    emoji: "🔥"
+---
+
+# Steel Browser
+
+Steel provides a REST API at \`http://{{STEEL_HOST}}:{{STEEL_PORT}}\` for AI agent web automation.
+
+## Create a Session
+
+\`\`\`bash
+curl -X POST http://{{STEEL_HOST}}:{{STEEL_PORT}}/v1/sessions \\
+  -H "Content-Type: application/json" \\
+  -d '{"blockAds": true}'
+\`\`\`
+
+## Scrape a Page
+
+\`\`\`bash
+curl -X POST http://{{STEEL_HOST}}:{{STEEL_PORT}}/v1/scrape \\
+  -H "Content-Type: application/json" \\
+  -d '{"url": "https://example.com", "format": "markdown"}'
+\`\`\`
+
+## Features
+
+- Session management with persistent cookies
+- Anti-detection and stealth plugins
+- Proxy support and IP rotation
+- Auto CAPTCHA solving
+- Puppeteer/Playwright/Selenium compatible
+`,
+};
+
+/**
+ * Generates SKILL.md files for each service that has skills defined.
+ *
+ * Returns a map of file paths (relative to project root) to file contents.
+ * Handlebars is used to replace `{{VAR}}` placeholders with actual values
+ * from each service's `openclawEnvVars`.
+ */
+export function generateSkillFiles(resolved: ResolverOutput): Record<string, string> {
+	const files: Record<string, string> = {};
+
+	for (const { definition } of resolved.services) {
+		if (definition.skills.length === 0) continue;
+
+		// Build a variable map from openclawEnvVars for Handlebars
+		const vars: Record<string, string> = {};
+		for (const envVar of definition.openclawEnvVars) {
+			// Resolve ${REFERENCES} to the default value (just strip the ${} wrapper)
+			const val = envVar.defaultValue.startsWith("${")
+				? envVar.defaultValue.slice(2, -1)
+				: envVar.defaultValue;
+			vars[envVar.key] = val;
+		}
+
+		for (const skill of definition.skills) {
+			const template = skillTemplates[skill.skillId];
+			if (!template) {
+				// Generate a generic skill file if no template exists
+				const generic = generateGenericSkill(
+					skill.skillId,
+					definition.name,
+					definition.icon,
+					definition.description,
+					vars,
+				);
+				files[`openclaw/workspace/skills/${skill.skillId}/SKILL.md`] = generic;
+				continue;
+			}
+
+			const compiled = Handlebars.compile(template, { noEscape: true });
+			const rendered = compiled(vars);
+			files[`openclaw/workspace/skills/${skill.skillId}/SKILL.md`] = rendered;
+		}
+	}
+
+	return files;
+}
+
+/**
+ * Generate a generic SKILL.md for skills that don't have a dedicated template.
+ */
+function generateGenericSkill(
+	skillId: string,
+	serviceName: string,
+	icon: string,
+	description: string,
+	vars: Record<string, string>,
+): string {
+	const title = skillId
+		.split("-")
+		.map((w) => w.charAt(0).toUpperCase() + w.slice(1))
+		.join(" ");
+
+	const envSection = Object.entries(vars)
+		.map(([key, value]) => `- **${key}:** \`${value}\``)
+		.join("\n");
+
+	return `---
+name: ${skillId}
+description: "${description}"
+metadata:
+  openclaw:
+    emoji: "${icon}"
+---
+
+# ${title}
+
+Interact with ${serviceName} through this skill.
+
+## Connection Details
+
+${envSection || "No specific connection variables configured."}
+
+## Usage Notes
+
+- This skill provides OpenClaw access to ${serviceName}.
+- Refer to the service documentation for available API endpoints.
+`;
+}