The context API for AI agents.
Entity search, indexed extract, and disruption watch for AI agents. Structured context with honest labels, never a silent fallback.
Try API for freeGround agents in owned sources
Over MCP, agents call entity search and crawl search_web:
passages from SeaWeb’s own index with the page URL and crawl date,
never a rented SERP silently filling gaps.
Models shouldn’t parse HTML
Hand over a URL and get the readable document back (title, clean text, passages). Indexed pages return with their crawl date. Free browser accounts stay index-only; Growth and API keys may live-fetch on a miss.
Structured context, or an honest no
Search responses are typed. When search_web can’t stand behind
an answer it says uncertain or unavailable, a label
your agent can act on, not a silent guess.
Example response
{
"coverage": "covered",
"results": [
{
"title": "Ferry Building Marketplace",
"url": "https://…/ferry-building",
"passage": "Open daily; ferry and
BART connections on the Embarcadero.",
"fetched_at": "2026-07-31",
"match_quality": "strong"
}
]
}Point your agent at it.
Try MCP with no key first: point your client at
api.seaweb.tech/mcp and send no Authorization header.
20 read calls per minute per IP.