Build with SearchPipe

Make your first search

Create an account, verify your email, then copy your default key from API Keys. Both integration paths use the same search pipeline and credit balance. The free plan provides 1,000 credits per UTC calendar month; sign in with a verified account each month to renew the allowance.

https://searchpipe.tech/mcp/?api_key=sp-YOUR_KEY

curl -X POST https://searchpipe.tech/v1/search \
  -H "Authorization: Bearer sp-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"FastAPI tutorial","max_results":5,"output_type":"answer"}'

import httpx

resp = httpx.post(
    "https://searchpipe.tech/v1/search",
    headers={"Authorization": "Bearer sp-YOUR_KEY"},
    json={"query": "FastAPI tutorial", "max_results": 5, "output_type": "answer"},
    timeout=60,
)
resp.raise_for_status()
data = resp.json()
print(data["answer"])
for r in data["results"]:
    print(r["score"], r["title"], r["url"])

const resp = await fetch("https://searchpipe.tech/v1/search", {
  method: "POST",
  headers: {
    "Authorization": "Bearer sp-YOUR_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ query: "FastAPI tutorial", max_results: 5, output_type: "answer" }),
});
if (!resp.ok) throw new Error(`Search failed (${resp.status}): ${await resp.text()}`);
const data = await resp.json();
console.log(data.answer);
for (const r of data.results) console.log(r.score, r.title, r.url);

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"io"
	"net/http"
	"time"
)

func main() {
	payload, _ := json.Marshal(map[string]any{
		"query":          "FastAPI tutorial",
		"max_results":    5,
		"output_type": "answer",
	})
	req, _ := http.NewRequest("POST", "https://searchpipe.tech/v1/search", bytes.NewReader(payload))
	req.Header.Set("Authorization", "Bearer sp-YOUR_KEY")
	req.Header.Set("Content-Type", "application/json")
	client := &http.Client{Timeout: 60 * time.Second}
	resp, err := client.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()
	body, _ := io.ReadAll(resp.Body)
	fmt.Println(string(body))
}

import java.net.URI;
import java.net.http.*;
import java.time.Duration;

public class SearchPipeDemo {
    public static void main(String[] args) throws Exception {
        String json = """
            {"query":"FastAPI tutorial","max_results":5,"output_type":"answer"}
            """;
        HttpClient client = HttpClient.newBuilder()
            .version(HttpClient.Version.HTTP_1_1)
            .connectTimeout(Duration.ofSeconds(10))
            .build();
        HttpRequest request = HttpRequest.newBuilder()
            .uri(URI.create("https://searchpipe.tech/v1/search"))
            .timeout(Duration.ofSeconds(60))
            .header("Authorization", "Bearer sp-YOUR_KEY")
            .header("Content-Type", "application/json")
            .POST(HttpRequest.BodyPublishers.ofString(json))
            .build();
        HttpResponse<String> resp =
            client.send(request, HttpResponse.BodyHandlers.ofString());
        System.out.println(resp.body());
    }
}

using System.Net.Http.Json;

using var client = new HttpClient();
client.DefaultRequestHeaders.Authorization =
    new System.Net.Http.Headers.AuthenticationHeaderValue("Bearer", "sp-YOUR_KEY");

var payload = new
{
    query = "FastAPI tutorial",
    max_results = 5,
    output_type = "answer",
};

HttpResponseMessage resp = await client.PostAsJsonAsync("https://searchpipe.tech/v1/search", payload);
resp.EnsureSuccessStatusCode();
Console.WriteLine(await resp.Content.ReadAsStringAsync());

Choose a search mode and output

search_depth controls processing and cost. output_type controls what you receive alongside the result list.

include_answer=true remains supported as a deprecated alias for output_type="answer". Explicit output_type always wins, including search_results.

Connect through MCP

Use https://searchpipe.tech/mcp/?api_key=sp-YOUR_KEY, or use https://searchpipe.tech/mcp/ with an Authorization: Bearer sp-YOUR_KEY header if your client supports it. Treat a URL containing your key as a credential.

claude mcp add --transport http searchpipe "https://searchpipe.tech/mcp/?api_key=sp-YOUR_KEY"

The search tool is searchpipe_search. It accepts the search parameters below, except stream. The searchpipe_verify tool supports separate claim verification. The searchpipe_account_balance tool returns the remaining credit balance and is free to call.

MCP setup & troubleshooting →Client configuration examples →

Search parameters

Send Content-Type: application/json. Only query is required. Start with the core fields; add filters and output options as needed.

Read the response

The result list is present in every output mode. Optional fields may be null. This illustrative basic answer response shows the envelope; values are examples, not a live search.

{
  "query": "FastAPI tutorial",
  "results": [{
    "id": "a43f294b8c6d012e",
    "url": "https://fastapi.tiangolo.com/tutorial/",
    "title": "Tutorial - User Guide - FastAPI",
    "content": "A step-by-step guide to using FastAPI.",
    "score": 0.94,
    "raw_content": null,
    "published_date": null,
    "evidence": null
  }],
  "answer": "See the FastAPI tutorial for a step-by-step guide. [0]",
  "ai_generated": true,
  "output": null,
  "answer_audit": null,
  "images": null,
  "request_id": "example-request-id",
  "response_time": 2.4,
  "usage": {"credits_charged": 1.0, "cached": false},
  "warnings": null,
  "notices": null
}

On basic/advanced, scores reflect LLM relevance ranking. Fast uses raw engine scores, which can exceed 1 and are not comparable across engines; use the returned order instead of applying a universal score threshold.

Evidence and audited answers

Search results can carry evidence without a separate verification call. Advanced generated answers can also include a statement-level audit. These are two different layers:

span_start is inclusive and span_end is exclusive. Offsets refer to extracted page text, or to content when evidence falls back to a search snippet. A snippet fallback is not a verified fetch of the original page. Request include_raw_content=true when you need the extracted text.

Source tiers are official, press, academic, preprint, community, or unknown. Fetch metadata includes fetch_status, redirected, final_url, and fetched_at. A source classification is not a truth verdict.

Audit statements include text, verdict, confidence, and citations. Verdicts are supported, partial, unsupported, or inconclusive. Statements without a valid citation marker are uncited and are not judged.

curl -X POST https://searchpipe.tech/v1/search \
  -H "Authorization: Bearer sp-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "What is retrieval-augmented generation?", "output_type": "answer", "search_depth": "advanced"}'

# response.answer_audit (nullable) -> score + per-statement verdicts
# response.results[n].evidence (nullable) -> quote, span, source tier

import httpx

resp = httpx.post(
    "https://searchpipe.tech/v1/search",
    headers={"Authorization": "Bearer sp-YOUR_KEY"},
    json={
        "query": "What is retrieval-augmented generation?",
        "output_type": "answer",
        "search_depth": "advanced",
    },
    timeout=60,
)
resp.raise_for_status()
data = resp.json()
print(data["answer"])
audit = data.get("answer_audit")
print("audit score:", audit["score"] if audit else "unavailable")
for s in (audit or {}).get("statements", []):
    print(s["verdict"], "-", s["text"])
for r in data["results"]:
    evidence = r.get("evidence")
    if evidence:
        print(evidence["quote"], "[", evidence["source_tier"], "]")

const resp = await fetch("https://searchpipe.tech/v1/search", {
  method: "POST",
  headers: {
    "Authorization": "Bearer sp-YOUR_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    query: "What is retrieval-augmented generation?",
    output_type: "answer",
    search_depth: "advanced",
  }),
});
if (!resp.ok) throw new Error(`Search failed (${resp.status}): ${await resp.text()}`);
const data = await resp.json();
console.log(data.answer);
console.log("audit score:", data.answer_audit?.score ?? "unavailable");
for (const s of data.answer_audit?.statements ?? []) console.log(s.verdict, "-", s.text);
for (const r of data.results) {
  if (r.evidence) console.log(r.evidence.quote, "[", r.evidence.source_tier, "]");
}

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"io"
	"net/http"
	"time"
)

func main() {
	payload, _ := json.Marshal(map[string]any{
		"query":          "What is retrieval-augmented generation?",
		"output_type": "answer",
		"search_depth":   "advanced",
	})
	req, _ := http.NewRequest("POST", "https://searchpipe.tech/v1/search", bytes.NewReader(payload))
	req.Header.Set("Authorization", "Bearer sp-YOUR_KEY")
	req.Header.Set("Content-Type", "application/json")
	client := &http.Client{Timeout: 60 * time.Second}
	resp, err := client.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()
	body, _ := io.ReadAll(resp.Body)
	fmt.Println(string(body)) // answer_audit.statements + results[n].evidence
}

import java.net.URI;
import java.net.http.*;
import java.time.Duration;

public class SearchPipeAuditedAnswer {
    public static void main(String[] args) throws Exception {
        String json = """
            {"query":"What is retrieval-augmented generation?","output_type":"answer","search_depth":"advanced"}
            """;
        HttpClient client = HttpClient.newBuilder()
            .version(HttpClient.Version.HTTP_1_1)
            .connectTimeout(Duration.ofSeconds(10))
            .build();
        HttpRequest request = HttpRequest.newBuilder()
            .uri(URI.create("https://searchpipe.tech/v1/search"))
            .timeout(Duration.ofSeconds(60))
            .header("Authorization", "Bearer sp-YOUR_KEY")
            .header("Content-Type", "application/json")
            .POST(HttpRequest.BodyPublishers.ofString(json))
            .build();
        HttpResponse<String> resp =
            client.send(request, HttpResponse.BodyHandlers.ofString());
        System.out.println(resp.body()); // answer_audit + evidence fields
    }
}

using System.Net.Http.Json;

using var client = new HttpClient();
client.DefaultRequestHeaders.Authorization =
    new System.Net.Http.Headers.AuthenticationHeaderValue("Bearer", "sp-YOUR_KEY");

var payload = new
{
    query = "What is retrieval-augmented generation?",
    output_type = "answer",
    search_depth = "advanced",
};

HttpResponseMessage resp = await client.PostAsJsonAsync("https://searchpipe.tech/v1/search", payload);
resp.EnsureSuccessStatusCode();
Console.WriteLine(await resp.Content.ReadAsStringAsync()); // answer_audit + evidence

Structured output

Set output_type="structured" and provide output_schema. SearchPipe generates a cited answer internally, then asks the model to convert it into a JSON object using your schema. Read the object from output; answer and answer_audit are null.

curl -X POST https://searchpipe.tech/v1/search \
  -H "Authorization: Bearer sp-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What is FastAPI and what language is it written in?",
    "output_type": "structured",
    "output_schema": {
      "type": "object",
      "properties": {
        "name": {"type": "string"},
        "language": {"type": "string"}
      },
      "required": ["name", "language"]
    }
  }'

Schemas must contain 1–20 flat properties of type string, number, integer, or boolean, and serialize to at most 4,000 characters. Nested objects and arrays are not supported. A missing or invalid schema returns 422.

Stream an answer

REST requests with stream=true return text/event-stream. Request an answer or structured output; streaming a results-only request returns 422.

curl -N -X POST https://searchpipe.tech/v1/search \
  -H "Authorization: Bearer sp-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"FastAPI tutorial","output_type":"answer","stream":true}'

event: meta
data: {"request_id":"example-id","query":"FastAPI tutorial","credits_charged":1.0}

event: done
data: {"query":"FastAPI tutorial","answer":null,"results":[],"ai_generated":false,"images":null,"answer_audit":null,"output":null,"request_id":"example-id","response_time":2.4,"usage":{"credits_charged":1.0,"cached":false},"warnings":["answer was requested but could not be generated (deadline or upstream failure)"],"notices":["no results matched the request"]}

Results or deltas can be absent, including on empty results or generation failure. Cache hits still use SSE, but delta counts can differ. Handle error events and a connection closing before done, even after HTTP 200. Client disconnection does not automatically refund a request.

Extract pages or verify a claim

Extract known URLs

POST /v1/extract retrieves extracted text for 1–20 public URLs (up to 2,048 characters each). It returns results with url/raw_content, and failed_results with url/error. Individual URL failures do not fail the whole batch. Private-network URLs are blocked.

curl -X POST https://searchpipe.tech/v1/extract \
  -H "Authorization: Bearer sp-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"urls":["https://fastapi.tiangolo.com/tutorial/"]}'

Default cost: 1 credit per URL. Failed URLs are refunded. Extract does not use the search-result cache or SearchResponse envelope.

Verify claims from your own pipeline Early access

For separately supplied claims and citations, use POST /v1/verify, POST /v1/verify/batch, or the MCP tool searchpipe_verify. Search evidence and advanced answer audits work within /search; they do not require this extra call.

Verification returns verdict, confidence, reason_code, and evidences. inconclusive is a valid billed outcome. Pipeline failures return 503 and are refunded.

Full verification contract and examples →

Official SDKs

The Python SDK is published on PyPI: pip install searchpipe. The TypeScript SDK is not yet published on npm; its source is available in the official integration repository. You can always use the HTTP examples for the current API contract.

from searchpipe import SearchPipe

client = SearchPipe(api_key="sp-YOUR_KEY")
response = client.search("FastAPI tutorial", include_answer=True)
print(response.answer)

import { SearchPipe } from "searchpipe";

const client = new SearchPipe({ apiKey: "sp-YOUR_KEY" });
const response = await client.search({ query: "FastAPI tutorial", includeAnswer: true });
console.log(response.answer);

Errors and retries

For REST API errors, inspect the HTTP status and error.code. Keep error.request_id when reporting a problem. The existing detail field is retained; for 422, it contains validation details.

{
  "detail": "Insufficient credits",
  "type": "error",
  "error": {
    "code": "insufficient_credits",
    "message": "Insufficient credits",
    "request_id": "example-request-id"
  }
}

A null answer/output or an empty result list can arrive with HTTP 200. Check the response before using it. Retrying is a new request and may incur a new charge.

Rate limits and billing

• Search credits: fast 0.5, basic 1, advanced 2 per call. Requesting an answer, images, raw content, or an available advanced audit does not add a separate search charge.
• Balance checks are free: GET /v1/account/balance (or the MCP tool searchpipe_account_balance) returns your remaining credits — not billed and not logged as usage.
• Cache hits are charged. The shared search cache avoids rerunning the pipeline, but does not waive the per-call credit cost. Read usage.credits_charged for the REST request.
• Refunds: pipeline failures and rejected output are automatically refunded. A successful empty result or unavailable optional answer is not automatically a failed billable request.
• Limits: the default is 100 requests per minute per API key, with a burst allowance of 20. REST and MCP share the same limit. On HTTP 429, follow Retry-After.
• Cache lifetime: default 300 seconds for matching search parameters. A cache hit still requires authentication, sufficient credits, and rate-limit checks.

Current pricing and credit plans → · View your usage →

More guides

MCP setup guideAgent setup guideFAQSearchPipe vs Tavily vs Exa vs BraveWeb Search APIs for AI AgentsWhat is a Web Search MCP server