OpenAPI spec
SerpKite's API is described by an OpenAPI 3.1 document: generate typed clients, import it into Postman, Insomnia or Bruno, or give it to an LLM.
The contract for api.serpkite.com is a single OpenAPI 3.1 file. It is the source of truth the API is built from: when an endpoint or field changes, this file changes first.
- Download: https://serpkite.com/openapi/serp-api.yaml
- Format: OpenAPI 3.1.0, YAML
- Contract version: 1.5.2 (the HTTP API remains
/v1)
curl -O https://serpkite.com/openapi/serp-api.yaml
What’s in it
- Servers:
https://api.serpkite.com. - Security scheme:
bearer(Authorization: Bearer skt_live_…). GET requests may also pass?api_key=. - Every endpoint:
/v1/search,/v1/images,/v1/videos,/v1/news,/v1/maps,/v1/places,/v1/place,/v1/reviews,/v1/shopping,/v1/scholar,/v1/patents,/v1/autocomplete,/v1/webpage,/v1/extract,/v1/map,/v1/crawl,/v1/crawl/{id},/v1/monitors,/v1/monitors/{id}, monitor/runsand/run,/v1/rank,POST /v1/batches,/v1/batches/{id},/v1/account,/v1/status,/customsearch/v1, plus the no-login/playground/searchdemo and the/v1/mcpendpoint. - Request schemas:
SearchRequest(shared by the search verticals),PlaceRequest,ReviewsRequest,WebpageRequest,ExtractRequest,MapRequest,CrawlRequest,MonitorCreateRequest,MonitorUpdateRequest,RankRequest,BatchCreateRequest. - Response schemas: one per search vertical (
SearchResponse,NewsResponse,PlacesResponsefor maps, places and place…), sharing therequest/results/metaenvelope (RequestEcho,Meta), plusWebpageResponse,ExtractResponse,MapResponse,TaskCreated,CrawlTask,Monitor,MonitorRunList,RankResponse,Batch,BatchCreateResponse,Account,Status,CSEResponseand theErrorshape. - Headers:
X-Credits-Used,X-Credits-Remaining,X-Cost-USD,X-Cache,X-Latency-MsandX-Tokens-Estimateon every billed response. - Alternate content types:
text/markdownresponses forformat: "markdown".
The dashboard backend (app-api.serpkite.com) is an internal API for the SerpKite dashboard and is not part of the public contract.
Site APIs and monitors have their own JSON envelopes. Search format: "markdown" returns a text body; extract and crawl format: "markdown" select Markdown content inside JSON. Follow the response schema for the endpoint rather than assuming every response has request, results and meta.
Generate a client
For TypeScript, Python and Go, use the official SDKs, whose TypeScript types and Python models are generated from this spec. SDK clients and Go types are maintained by hand, with contract checks for Go. For other languages, or if you want your own types, generators produce good clients from the spec.
TypeScript types with openapi-typescript, used with openapi-fetch:
npx openapi-typescript https://serpkite.com/openapi/serp-api.yaml -o src/serpkite.d.ts
Go with oapi-codegen:
go install github.com/oapi-codegen/oapi-codegen/v2/cmd/oapi-codegen@latest
curl -sO https://serpkite.com/openapi/serp-api.yaml
oapi-codegen -generate types,client -package serpkite serp-api.yaml > serpkite/client.gen.go
Python, Java, C#, Ruby, PHP and more with OpenAPI Generator:
npx @openapitools/openapi-generator-cli generate \
-i https://serpkite.com/openapi/serp-api.yaml \
-g python \
-o ./serpkite-client
Some generators struggle with OpenAPI 3.1 oneOf unions, such as the job-or-error entries of BatchCreateResponse or the engine-or-list engine parameter. If one does, keep the generated models and send requests with your own thin wrapper.
Import into an API client
- Postman: Import → Link, paste
https://serpkite.com/openapi/serp-api.yaml. Set the collection’s auth to Bearer Token with your key. - Insomnia: Create → Import → From URL, paste the same URL.
- Bruno: Import Collection → OpenAPI V3 Spec, choose the downloaded file.
Keep the key in an environment variable of the tool, not in the saved requests you share.
Use it with an LLM
The spec is compact enough to paste into a model’s context when you ask it to write integration code. For the prose docs as well, use llms.txt (an index) or llms-full.txt (every docs page as Markdown), or add .md to any docs URL.
Versioning
The HTTP API uses /v1; the OpenAPI document currently has contract version 1.5.2. Additive changes can update the contract version without changing the URL prefix. Parse responses tolerantly and ignore unknown fields. Breaking changes will get a new version and advance notice in the API changelog.
Last updated: 2026-10-21