What changed

research-topic-scan runs one more search leg: x-search. A single query now fans out to web-search, news-search, wikipedia-search, arxiv-search, hn-search, and x-search in parallel, merged into one deduped result list. X/Twitter results carry source: "x" and link straight to the post. 

{
  "query": "agentic commerce x402 protocol"
}

returns results with an added source in the mix: 

{
  "by_source": { "web": 4, "news": 3, "wikipedia": 1, "arxiv": 2, "hn": 2, "x": 1 },
  "results": [
    {
      "title": "x402 agent payments are picking up",
      "url": "https://x.com/agentutility_ai/status/1234567890",
      "source": "x",
      "published_at": "2026-07-11T00:00:00Z"
    }
  ]
}

Why

research-topic-scan was built to replace the five separate lookups an agent would otherwise run to scope a topic: web, news, an encyclopedia entry, academic papers, and developer chatter. X/Twitter chatter is the same kind of signal — recent public discussion of a topic — and buyers pairing arxiv-search or serp-search with twitter-search on their own were already doing this fan-out by hand.

Degradation

x-search is optional, same as news, wikipedia, arxiv, and hn: only web-search is required. If the X API call fails or the endpoint isn't configured with a bearer token, research-topic-scan still returns whatever the other legs found and marks the response degraded rather than failing the whole call.

Price stays at $0.06.