Developer reference
SearchPipe API overview
SearchPipe is a web search API and MCP server for developers building AI agents. Choose an endpoint below, copy a request, then open the reference for details.
REST APIs at a glance
All four endpoints use JSON and an API key. The versioned paths below and their original aliases share the same behavior and billing.
| Endpoint | Use & key input | Returns | Credits |
|---|---|---|---|
POST/v1/searchSearch reference → |
Find sources for a question or topic, with optional cited answers and structured output.query; optional search_depth, output_type, source and date filters. | Ranked results, plus an optional answer or output object, evidence and advanced answer audits. | fast: 0.5 · basic: 1 · advanced: 2 credits / call |
POST/v1/extractExtract schema → |
Fetch readable text from URLs you already know, without running a search or generating an answer.urls: an array of 1–20 public URLs. | results with extracted raw_content, and failed_results with per-URL errors. | Default: 1 credit / URL; failed URLs refunded |
POST/v1/verifyEarly accessVerification reference → |
Audit a claim from your own pipeline against retrieved evidence or citations you supply.claim; optional mode, evidence_urls and max_evidence. | A verdict, confidence, reason_code, evidence items and timing. | Default: 1 credit / claim; inconclusive is charged |
POST/v1/verify/batchEarly accessBatch schema → |
Check up to 20 claims in one request and handle each result independently.claims: 1–20 statements; optional shared mode and max_evidence. | Ordered results, with null for failed items, and a failed array identifying errors by index. | Default: 1 credit / claim; failed items refunded |
Credits are shared across REST and MCP. See plans and credit prices for the cost of credits.
Before your first request
Create an account, verify your email, and copy a key from API Keys. Send Authorization: Bearer sp-YOUR_KEY or X-API-Key: sp-YOUR_KEY with Content-Type: application/json. Replace the placeholder in each example with your key.
Common failures: 401 invalid key, 403 unverified email, 402 insufficient credits, 422 invalid input, 429 rate limit. Error handling and retries →
Search the web
Find sources for a question or topic, with optional cited answers and structured output.
/v1/searchAlias: /searchKey inputs
query- Required search question or keywords.
search_depth- fast (0.5 credits), basic (1 credit, default), or advanced (2 credits).
output_type- search_results (default), answer, or structured. structured also requires output_schema.
What you receive
results- Source URLs, titles, content and scores; evidence is included when available.
answer / output- Optional cited answer or JSON object. Fast skips all LLM steps and returns results only.
answer_audit- Statement-level audit for advanced answer mode when available; may be null.
usage- Credits charged and cache status. Cache hits are charged at the same depth rate.
- Use stream=true for Server-Sent Events with answer or structured output. Streaming is available through REST only.
- See the search reference for every parameter, output limitations and response fields.
Copy a curl request
curl -X POST "https://searchpipe.tech/v1/search" \
-H "Authorization: Bearer sp-YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"query": "FastAPI tutorial",
"search_depth": "basic",
"output_type": "answer",
"max_results": 5
}'Extract page content
Fetch readable text from URLs you already know, without running a search or generating an answer.
/v1/extractAlias: /extractKey inputs
urls- Required array of 1–20 URLs; each URL is limited to 2,048 characters. Only public HTTP(S) URLs are allowed.
What you receive
results[]- Successful items contain url and raw_content (extracted text, not HTML).
failed_results[]- Failed items contain url and error. A partial or fully failed extraction can still return HTTP 200; inspect this array.
- Credits are deducted before fetching. Failed URLs are refunded; if every URL fails, the whole charge is refunded.
- Extract does not run LLM summarization or reranking and does not use the search result cache.
Copy a curl request
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/"
]
}'Verify a claim
Early accessAudit a claim from your own pipeline against retrieved evidence or citations you supply.
/v1/verifyAlias: /verifyKey inputs
claim- Required statement to verify; 1–2,000 characters.
mode- auto (default), paper, or web. auto selects the evidence route using DOI or arXiv signals.
evidence_urls- Optional list of up to 5 citation URLs to check. Available on single-claim verification only.
max_evidence- 1–10 evidence items; default 5.
What you receive
verdict- supported, partial, unsupported, or inconclusive.
confidence / reason_code- Confidence from 0 to 1 and a machine-readable reason for the verdict.
evidences[]- Source URLs with available quotes, span offsets, source tiers and per-evidence verdicts. Optional fields may be null.
timing- Processing time for individual verification stages, in seconds.
- An inconclusive verdict is a completed verification and is charged. Overall service failures are refunded.
- This endpoint is in early access. For SearchPipe-generated answers, start with advanced search and its built-in answer_audit.
Copy a curl request
curl -X POST "https://searchpipe.tech/v1/verify" \
-H "Authorization: Bearer sp-YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"claim": "FastAPI is a Python web framework.",
"mode": "web",
"evidence_urls": [
"https://fastapi.tiangolo.com/"
],
"max_evidence": 5
}'Verify a batch of claims
Early accessCheck up to 20 claims in one request and handle each result independently.
/v1/verify/batchAlias: /verify/batchKey inputs
claims- Required array of 1–20 statements; each is limited to 2,000 characters.
mode- auto (default), paper, or web; applied to the entire batch.
max_evidence- 1–10 evidence items per claim; default 5. This endpoint does not accept evidence_urls.
What you receive
results[]- Matches the order of claims. Each item has the single-claim response shape, or null when that item fails.
failed[]- Each failure includes index (zero-based position in claims) and error. Partial failures do not fail the whole batch.
- The batch is charged before execution. Failed items are refunded; completed inconclusive verdicts are charged.
- This is a synchronous REST endpoint, not a background job. There is no separate MCP batch tool.
Copy a curl request
curl -X POST "https://searchpipe.tech/v1/verify/batch" \
-H "Authorization: Bearer sp-YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"claims": [
"FastAPI is a Python web framework.",
"Python was created by Guido van Rossum."
],
"mode": "web",
"max_evidence": 5
}'Using an MCP client?
Connect the same account and credit balance to Claude Code, Cursor, Codex or another MCP client. The remote MCP server exposes these three tools:
| MCP tool | REST equivalent | Scope |
|---|---|---|
searchpipe_search | POST /v1/search | Search parameters and output modes; stream is REST only. |
searchpipe_verify | POST /v1/verify | Single-claim verification, including optional citation URLs. Early access. |
searchpipe_account_balance | GET /v1/account/balance | Remaining credits and next expiry; free to call. |
Extraction and batch verification are available through REST. MCP uses Streamable HTTP transport; this is separate from REST search SSE output.
Schemas, management APIs & health
This page covers the APIs used by agents to search, extract and verify. Agents can check remaining credits programmatically with GET /v1/account/balance using the same API key — it is free to call. The remaining account, API key, billing and usage management interfaces require signing in to the console; they are listed in the complete OpenAPI reference.