OpenClaw CLI + Agents
Connect OpenClaw to Orsa so your OpenClaw agents can scrape pages, extract products, run brand lookups, and query websites with real-time context.
Fastest setup (from the dashboard)
- Open the Orsa dashboard sidebar.
- Click OpenClaw (red lobster CTA at the bottom).
- Click Generate Key + Copy Prompt.
- Paste that prompt directly into OpenClaw.
That flow creates a fresh API key named OpenClaw and gives you a ready-to-send prompt that includes:
- your key
- Orsa base URL
- common API endpoints
- MCP configuration
Playground template catalog
The dashboard playground now supports template-first request building with category filters and Request / Response / Component tabs.
Categories
- Bing
- Walmart
- Target
- Amazon
- Universal
- TikTok
- YouTube
- AI Tools
- Brand
Scraping templates
- Google Search with AI Overview
- Amazon Search
- Web
- Google Travel Hotels
- Google Ads with AI Overview
- Google Lens
- Bing
- Bing Search
- Walmart Product
- Walmart Search
- Walmart
- Target
- Target Product
- Target Search
- Amazon
- Amazon Product
- Amazon Pricing
- Amazon Sellers
- Amazon Bestsellers
- TikTok Post
- TikTok Shop Search
- TikTok Shop Product
- TikTok Shop URL
- Reddit Post
- Reddit Subreddit
- Reddit User
- YouTube Metadata
- YouTube Search
- YouTube Subtitles
- YouTube Channel
- ChatGPT
- Perplexity
- Google AI Mode
Brand templates
- Brand Retrieve
- Brand Screenshot
- Brand AI Products
- Brand AI Product
- Brand AI Query
OpenClaw environment variables
If you prefer manual setup, add these env vars in OpenClaw:
ORSA_API_KEY=or_live_xxxxxxxxxxxxxxxxxxxxx
ORSA_BASE_URL=https://api.orsa.devOpenClaw MCP configuration
Use the Orsa MCP server so OpenClaw agents can call Orsa tools directly.
{
"mcpServers": {
"orsa": {
"command": "npx",
"args": ["@orsa-dev/mcp-server"],
"env": {
"ORSA_API_KEY": "or_live_xxxxxxxxxxxxxxxxxxxxx",
"ORSA_BASE_URL": "https://api.orsa.dev"
}
}
}
}CLI-first API examples
Scrape markdown
curl -X GET "https://api.orsa.dev/api/v1/web/scrape/markdown?url=https://stripe.com/pricing" \
-H "Authorization: Bearer $ORSA_API_KEY"Retrieve brand data
curl -X GET "https://api.orsa.dev/api/v1/brand/retrieve?domain=linear.app" \
-H "Authorization: Bearer $ORSA_API_KEY"AI extraction query
curl -X POST "https://api.orsa.dev/api/v1/brand/ai/query" \
-H "Authorization: Bearer $ORSA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"domain": "stripe.com",
"query": "return every pricing tier with price and notable features"
}'Agent prompt starter for OpenClaw
Use this if you are not using the in-app generator:
Use Orsa as your web context tool for this workspace.
ORSA_BASE_URL=https://api.orsa.dev
ORSA_API_KEY=<insert key>
Prefer these tools first:
1) web scrape markdown for live content retrieval
2) brand retrieve for company context
3) AI query when schema-free extraction is needed
4) products endpoint for ecommerce and catalog intelligence
If MCP is available, use @orsa-dev/mcp-server with ORSA_API_KEY for tool access.Tool map for OpenClaw agents
| Need | Endpoint | Credit cost |
|---|---|---|
| Web content (LLM-ready) | GET /api/v1/web/scrape/markdown | 1 |
| Raw HTML | GET /api/v1/web/scrape/html | 1 |
| Images on page | GET /api/v1/web/scrape/images | 1 |
| Brand profile | GET /api/v1/brand/retrieve | 10 |
| AI query | POST /api/v1/brand/ai/query | 20 |
| Product extraction | GET /api/v1/brand/ai/products | 15 |
| Single product extraction | GET /api/v1/brand/ai/product | 10 |
Best practices
- Generate one dedicated key per OpenClaw workspace for easier rotation.
- Set per-agent credit budgets to prevent runaway autonomous loops.
- Prefer markdown scraping for lower token usage and cleaner context.
- Cache repeated domain calls in your agent memory/store.
- Rotate keys immediately if a workspace or prompt transcript is exposed.