Where Lookout sits

Lookout gives an agent search hits and fetched excerpts it can cite, without an API key. It is not a hosted search API and it does not render pages in a browser.

The everyday calls are web_search, web_fetch, and web_extract. web_research is a separate multi-step tool. It is not what search does.

Peers

Peer	What they are	What Lookout does
Firecrawl	A hosted crawl and scrape API. You send a key. It can return page content as markdown and can run a browser.	Lookout runs as a stdio process next to the agent. web_fetch reads the HTTP body and stops at the size, time, and redirect caps. It does not execute JavaScript.
Tavily	A hosted search API for agents. A key returns results and can include extracted page content.	web_search does not call Tavily. It queries DuckDuckGo HTML, Wikipedia OpenSearch, the npm registry, and Hacker News Algolia, and it names the adapter on each hit. Opening a page is web_fetch or web_extract.
wigolo	An MCP server for search, fetch, crawl, and research. Its repository says that path does not require an API key.	Lookout also searches and fetches without an API key. Crawl and multi-step research are tools you call by name. web_search does not crawl and does not fetch the result pages.

No latency comparison is claimed. A failed adapter stays a warning.

Not the job

• A required paid search API.
• Browser rendering or a model-written summary as the evidence.
• Treating web_research as the default call.

Install

npx -y @sylphx/lookout

Bare invoke starts the MCP stdio server. The same package serves the CLI subcommands.