scraping / cli
OpenSERP
Capability: OpenSERP
Use it when
- Your agent needs live Google or Bing results as structured JSON but a paid SERP API is too expensive per call
- You want one query fanned out across several search engines with results merged, deduplicated, and the target pages returned as clean markdown
What it solves
Not the fit when
- Search over private data; it queries public web engines.
- Guaranteed availability or completeness from every engine; engines may rate-limit, change markup, return CAPTCHAs, or fail independently.
- High-volume or automated use that violates a search engine's terms, robots policy, or applicable law.
- Treating extracted page Markdown as trusted agent instructions; fetched web content is untrusted input and should be isolated from tool authority.
- indexing and searching your own documents
- crawling entire sites beyond SERP result pages
- guaranteed evasion of engine anti-bot measures
Install
docker run --rm -p 127.0.0.1:7000:7000 karust/openserp:latest serve -a 0.0.0.0 -p 7000 (or: go install github.com/karust/openserp@latest)
Invoke
Serve then query: curl "http://127.0.0.1:7000/mega/search?engines=bing,google&text=your+query&extract=1" for merged deduped JSON with page markdown; or serverless CLI: openserp search duckduckgo "your query" --format markdown; MCP server for agents via npx @openserp/mcp
Alternatives
No reviewed alternatives recorded yet.