Web access for AI agentsCitedScoredCached

The Web Gateway your agents can trust

Search the open web via SearXNG, then return clean JSON: de-duped URLs, trust & recency scores, and RAG answers with citations.

Reliable sources

Trust scoring + allow/block policies.

Keep your agent grounded: prefer reputable domains, penalize spammy patterns, expose score reasons.

Agent-grade output

Stable schemas, tool-friendly JSON.

Designed for LangChain/CrewAI tools: fast search, async RAG jobs, citations you can inspect.

Built to ship

Cache + credits transparency.

Cache results when it makes sense, bypass when you want freshness. Every response returns `credits_used`.
Live demo
WebGateway
Search
Trust score
0.82
Latency
612ms
One call
/v1/search
curl -X POST $API/v1/search \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "best ev battery 2025",
    "categories": ["web"],
    "limit": 20,
    "max_pages": 5,
    "cache": "bypass",
    "source_policy": "allowlist",
    "domains": ["iea.org","npr.org"]
  }'
Response highlights
  • • `meta.unique_results` + deep paging stats
  • • `scores.final` + score reasons
  • • `meta.unresponsive_engines` for CAPTCHAs

Single tool call

Drop-in “web gateway” for your agent runtime.

curl -X POST $API/v1/search \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "best ev battery 2025",
    "limit": 20,
    "max_pages": 5,
    "cache": "bypass"
  }'

Structured response

Scores included, citations when in RAG mode.

{
  "results": [
    {
      "url": "...",
      "scores": { "final": 0.79, "trust": 0.82, "recency": 0.71 }
    }
  ],
  "meta": { "credits_used": 1 }
}

Ship web-enabled agents, safely.

You’re not buying “search results”. You’re buying a web intelligence layer: caching, scoring, and citations you can trust.