openserp/docs/openapi.yaml

openapi: 3.0.3
info:
  title: OpenSERP API
  version: 2.2.0
  description: >
    OpenSERP provides dedicated and multi-engine search endpoints for Google, Yandex,
    Baidu, Bing, and DuckDuckGo. Search responses are wrapped in a v2 envelope with
    query echo, metadata, normalized results, and pagination. Invalid client input
    returns 400 with a machine-readable `reason` code.
  license:
    name: MIT
    url: https://opensource.org/licenses/MIT
servers:
  - url: http://127.0.0.1:7000
    description: Local default server
security: []
tags:
  - name: Search
    description: Dedicated per-engine search endpoints
  - name: Mega
    description: Cross-engine aggregated search endpoints
  - name: Health
    description: Health and readiness endpoints
  - name: Stats
    description: Runtime statistics endpoints
  - name: Docs
    description: OpenAPI and Swagger UI endpoints
paths:
  /{engine}/search:
    get:
      tags: [Search]
      operationId: searchWeb
      summary: Search web results from a specific engine
      description: >
        Engine path values are `google`, `yandex`, `baidu`, `bing`, `duck`, and `ecosia`
        (`duck` maps to DuckDuckGo internally). Use `?format=markdown|text|ndjson`
        for alternative output formats.
      parameters:
        - $ref: "#/components/parameters/EnginePath"
        - $ref: "#/components/parameters/TextQuery"
        - $ref: "#/components/parameters/LangQuery"
        - $ref: "#/components/parameters/RegionQuery"
        - $ref: "#/components/parameters/DateQuery"
        - $ref: "#/components/parameters/FileQuery"
        - $ref: "#/components/parameters/SiteQuery"
        - $ref: "#/components/parameters/LimitQuery"
        - $ref: "#/components/parameters/StartQuery"
        - $ref: "#/components/parameters/FilterQuery"
        - $ref: "#/components/parameters/FeaturesQuery"
        - $ref: "#/components/parameters/ExtractQuery"
        - $ref: "#/components/parameters/ExtractModeQuery"
        - $ref: "#/components/parameters/MinRunesQuery"
        - $ref: "#/components/parameters/FormatQuery"
        - $ref: "#/components/parameters/UseProxyHeader"
        - $ref: "#/components/parameters/ProxyURLHeader"
        - $ref: "#/components/parameters/ProxyCountryHeader"
        - $ref: "#/components/parameters/ProxyClassHeader"
        - $ref: "#/components/parameters/ProxyProviderHeader"
        - $ref: "#/components/parameters/ProxySessionIDHeader"
        - $ref: "#/components/parameters/TenantHeader"
      responses:
        "200":
          description: Search results envelope
          headers:
            X-Request-ID:
              $ref: "#/components/headers/XRequestID"
            X-Cache:
              $ref: "#/components/headers/XCache"
            X-Fallback-Engine:
              $ref: "#/components/headers/XFallbackEngine"
            X-Proxy-Mode:
              $ref: "#/components/headers/XProxyMode"
            X-Proxy-Tag:
              $ref: "#/components/headers/XProxyTag"
            X-Proxy-Used:
              $ref: "#/components/headers/XProxyUsed"
            X-Network-Bytes:
              $ref: "#/components/headers/XNetworkBytes"
            X-Browser-Profile-Id:
              $ref: "#/components/headers/XBrowserProfileID"
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/SearchEnvelope"
              examples:
                default:
                  value:
                    query:
                      text: golang
                      lang: EN
                      region: US
                      engines_requested: [google]
                    meta:
                      request_id: 01HXYZ...
                      requested_at: "2026-04-24T12:00:00Z"
                      took_ms: 842
                      engines_failed: []
                      version: "2.1"
                    results:
                      - id: s_a1b2c3d4e5f6a1b2
                        rank: 1
                        type: organic
                        title: The Go Programming Language
                        url: https://go.dev/
                        display_url: go.dev
                        snippet: Go is an open source programming language...
                        domain: go.dev
                        favicon: https://go.dev/favicon.ico
                        position:
                          absolute: 1
                        engine: google
                        domain_info:
                          tld: dev
                          sld: go
                          category: ""
                    serp_features:
                      - id: f_a1b2c3d4e5f6a1b2
                        engine: google
                        type: ai_summary
                        text: Go is an open source programming language used for fast, reliable services.
                        links:
                          - title: The Go Programming Language
                            url: https://go.dev/
                        source_result_ids: [s_a1b2c3d4e5f6a1b2]
                        position:
                          absolute: 1
                        confidence: 0.95
                        extracted_at: "2026-04-24T12:00:00Z"
                    pagination:
                      page: 1
                      has_more: true
                      next_start: 25
            text/markdown:
              schema:
                type: string
            text/plain:
              schema:
                type: string
            application/x-ndjson:
              schema:
                type: string
        "400":
          $ref: "#/components/responses/BadRequestError"
        "403":
          $ref: "#/components/responses/ForbiddenError"
        "429":
          $ref: "#/components/responses/TooManyRequestsError"
        "502":
          $ref: "#/components/responses/BadGatewayError"
        "503":
          $ref: "#/components/responses/ServiceUnavailableError"
        "504":
          $ref: "#/components/responses/GatewayTimeoutError"
        "404":
          $ref: "#/components/responses/NotFoundError"
        "500":
          $ref: "#/components/responses/InternalServerError"
  /{engine}/image:
    get:
      tags: [Search]
      operationId: searchImages
      summary: Search image results from a specific engine
      parameters:
        - $ref: "#/components/parameters/EnginePath"
        - $ref: "#/components/parameters/TextQuery"
        - $ref: "#/components/parameters/LangQuery"
        - $ref: "#/components/parameters/RegionQuery"
        - $ref: "#/components/parameters/DateQuery"
        - $ref: "#/components/parameters/FileQuery"
        - $ref: "#/components/parameters/SiteQuery"
        - $ref: "#/components/parameters/LimitQuery"
        - $ref: "#/components/parameters/StartQuery"
        - $ref: "#/components/parameters/FilterQuery"
        - $ref: "#/components/parameters/FeaturesQuery"
        - $ref: "#/components/parameters/FormatQuery"
        - $ref: "#/components/parameters/UseProxyHeader"
        - $ref: "#/components/parameters/ProxyURLHeader"
        - $ref: "#/components/parameters/ProxyCountryHeader"
        - $ref: "#/components/parameters/ProxyClassHeader"
        - $ref: "#/components/parameters/ProxyProviderHeader"
        - $ref: "#/components/parameters/ProxySessionIDHeader"
        - $ref: "#/components/parameters/TenantHeader"
      responses:
        "200":
          description: Image search results envelope
          headers:
            X-Request-ID:
              $ref: "#/components/headers/XRequestID"
            X-Cache:
              $ref: "#/components/headers/XCache"
            X-Fallback-Engine:
              $ref: "#/components/headers/XFallbackEngine"
            X-Proxy-Mode:
              $ref: "#/components/headers/XProxyMode"
            X-Proxy-Tag:
              $ref: "#/components/headers/XProxyTag"
            X-Proxy-Used:
              $ref: "#/components/headers/XProxyUsed"
            X-Network-Bytes:
              $ref: "#/components/headers/XNetworkBytes"
            X-Browser-Profile-Id:
              $ref: "#/components/headers/XBrowserProfileID"
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ImageEnvelope"
        "400":
          $ref: "#/components/responses/BadRequestError"
        "403":
          $ref: "#/components/responses/ForbiddenError"
        "429":
          $ref: "#/components/responses/TooManyRequestsError"
        "502":
          $ref: "#/components/responses/BadGatewayError"
        "503":
          $ref: "#/components/responses/ServiceUnavailableError"
        "504":
          $ref: "#/components/responses/GatewayTimeoutError"
        "404":
          $ref: "#/components/responses/NotFoundError"
        "500":
          $ref: "#/components/responses/InternalServerError"
  /google/parse:
    post:
      tags: [Search]
      operationId: parseGoogleHTML
      summary: Parse a Google SERP HTML document into structured results
      description: >
        Accepts raw Google SERP HTML in the request body and returns a standard
        search envelope. Useful when an upstream provider delivers raw HTML
        rather than JSON. No browser is used; parsing is done with goquery.
        The body size limit is 10 MB.
      requestBody:
        required: true
        content:
          text/html:
            schema:
              type: string
              description: Raw Google SERP HTML page
      parameters:
        - $ref: "#/components/parameters/FormatQuery"
      responses:
        "200":
          description: Parsed search results envelope
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/SearchEnvelope"
            text/markdown:
              schema:
                type: string
            text/plain:
              schema:
                type: string
            application/x-ndjson:
              schema:
                type: string
        "400":
          $ref: "#/components/responses/BadRequestError"
        "500":
          $ref: "#/components/responses/InternalServerError"
  /bing/parse:
    post:
      tags: [Search]
      operationId: parseBingHTML
      summary: Parse a Bing SERP HTML document into structured results
      description: >
        Accepts raw Bing SERP HTML in the request body and returns a standard
        search envelope. Useful when an upstream provider delivers raw HTML
        rather than JSON. No browser is used; parsing is done with goquery.
        The body size limit is 10 MB.
      requestBody:
        required: true
        content:
          text/html:
            schema:
              type: string
              description: Raw Bing SERP HTML page
      parameters:
        - $ref: "#/components/parameters/FormatQuery"
      responses:
        "200":
          description: Parsed search results envelope
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/SearchEnvelope"
            text/markdown:
              schema:
                type: string
            text/plain:
              schema:
                type: string
            application/x-ndjson:
              schema:
                type: string
        "400":
          $ref: "#/components/responses/BadRequestError"
        "500":
          $ref: "#/components/responses/InternalServerError"
  /mega/search:
    get:
      tags: [Mega]
      operationId: megaSearch
      summary: Search across multiple engines with selectable execution mode
      description: >
        Mode controls engine execution strategy: `balanced` (default) queries all
        selected engines in parallel, `any` runs engines sequentially in requested
        order until first success, and `fast` queries only the fastest engine based
        on circuit-breaker average response time stats.
        In `balanced` mode, `dedupe` and `merge` tune aggregation behavior.
        Partial failures are surfaced in `meta.engines_failed` and
        `meta.engine_errors`. If all selected engines fail, the endpoint returns
        a 502 with per-engine error details. Use `?format=markdown|text|ndjson`
        for alternative output formats.
      parameters:
        - $ref: "#/components/parameters/TextQuery"
        - $ref: "#/components/parameters/LangQuery"
        - $ref: "#/components/parameters/RegionQuery"
        - $ref: "#/components/parameters/DateQuery"
        - $ref: "#/components/parameters/FileQuery"
        - $ref: "#/components/parameters/SiteQuery"
        - $ref: "#/components/parameters/LimitQuery"
        - $ref: "#/components/parameters/StartQuery"
        - $ref: "#/components/parameters/FilterQuery"
        - $ref: "#/components/parameters/FeaturesQuery"
        - $ref: "#/components/parameters/EnginesQuery"
        - $ref: "#/components/parameters/MegaModeQuery"
        - $ref: "#/components/parameters/MegaDedupeQuery"
        - $ref: "#/components/parameters/MegaMergeQuery"
        - $ref: "#/components/parameters/ExtractQuery"
        - $ref: "#/components/parameters/ExtractModeQuery"
        - $ref: "#/components/parameters/MinRunesQuery"
        - $ref: "#/components/parameters/FormatQuery"
        - $ref: "#/components/parameters/UseProxyHeader"
        - $ref: "#/components/parameters/ProxyURLHeader"
        - $ref: "#/components/parameters/ProxyCountryHeader"
        - $ref: "#/components/parameters/ProxyClassHeader"
        - $ref: "#/components/parameters/ProxyProviderHeader"
        - $ref: "#/components/parameters/ProxySessionIDHeader"
        - $ref: "#/components/parameters/TenantHeader"
      responses:
        "200":
          description: Aggregated envelope with clusters
          headers:
            X-Request-ID:
              $ref: "#/components/headers/XRequestID"
            X-Cache:
              $ref: "#/components/headers/XCache"
            X-Proxy-Mode:
              $ref: "#/components/headers/XProxyMode"
            X-Proxy-Tag:
              $ref: "#/components/headers/XProxyTag"
            X-Proxy-Used:
              $ref: "#/components/headers/XProxyUsed"
            X-Network-Bytes:
              $ref: "#/components/headers/XNetworkBytes"
            X-Browser-Profile-Id:
              $ref: "#/components/headers/XBrowserProfileID"
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/MegaSearchEnvelope"
            text/markdown:
              schema:
                type: string
            text/plain:
              schema:
                type: string
            application/x-ndjson:
              schema:
                type: string
        "400":
          $ref: "#/components/responses/BadRequestError"
        "403":
          $ref: "#/components/responses/ForbiddenError"
        "429":
          $ref: "#/components/responses/TooManyRequestsError"
        "502":
          $ref: "#/components/responses/BadGatewayError"
        "503":
          $ref: "#/components/responses/ServiceUnavailableError"
        "504":
          $ref: "#/components/responses/GatewayTimeoutError"
        "500":
          $ref: "#/components/responses/InternalServerError"
  /mega/image:
    get:
      tags: [Mega]
      operationId: megaImageSearch
      summary: Image search across multiple engines with selectable execution mode
      parameters:
        - $ref: "#/components/parameters/TextQuery"
        - $ref: "#/components/parameters/LangQuery"
        - $ref: "#/components/parameters/RegionQuery"
        - $ref: "#/components/parameters/DateQuery"
        - $ref: "#/components/parameters/FileQuery"
        - $ref: "#/components/parameters/SiteQuery"
        - $ref: "#/components/parameters/LimitQuery"
        - $ref: "#/components/parameters/StartQuery"
        - $ref: "#/components/parameters/FilterQuery"
        - $ref: "#/components/parameters/FeaturesQuery"
        - $ref: "#/components/parameters/EnginesQuery"
        - $ref: "#/components/parameters/MegaModeQuery"
        - $ref: "#/components/parameters/MegaDedupeQuery"
        - $ref: "#/components/parameters/MegaMergeQuery"
        - $ref: "#/components/parameters/FormatQuery"
        - $ref: "#/components/parameters/UseProxyHeader"
        - $ref: "#/components/parameters/ProxyURLHeader"
        - $ref: "#/components/parameters/ProxyCountryHeader"
        - $ref: "#/components/parameters/ProxyClassHeader"
        - $ref: "#/components/parameters/ProxyProviderHeader"
        - $ref: "#/components/parameters/ProxySessionIDHeader"
        - $ref: "#/components/parameters/TenantHeader"
      responses:
        "200":
          description: Aggregated image results envelope
          headers:
            X-Request-ID:
              $ref: "#/components/headers/XRequestID"
            X-Cache:
              $ref: "#/components/headers/XCache"
            X-Proxy-Mode:
              $ref: "#/components/headers/XProxyMode"
            X-Proxy-Tag:
              $ref: "#/components/headers/XProxyTag"
            X-Proxy-Used:
              $ref: "#/components/headers/XProxyUsed"
            X-Network-Bytes:
              $ref: "#/components/headers/XNetworkBytes"
            X-Browser-Profile-Id:
              $ref: "#/components/headers/XBrowserProfileID"
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ImageEnvelope"
        "400":
          $ref: "#/components/responses/BadRequestError"
        "403":
          $ref: "#/components/responses/ForbiddenError"
        "429":
          $ref: "#/components/responses/TooManyRequestsError"
        "502":
          $ref: "#/components/responses/BadGatewayError"
        "503":
          $ref: "#/components/responses/ServiceUnavailableError"
        "504":
          $ref: "#/components/responses/GatewayTimeoutError"
        "500":
          $ref: "#/components/responses/InternalServerError"
  /mega/engines:
    get:
      tags: [Mega]
      operationId: listMegaEngines
      summary: List available engines and runtime state
      responses:
        "200":
          description: Engine list
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/MegaEnginesResponse"
  /extract:
    get:
      tags: [Search]
      operationId: extractURL
      summary: Extract clean content from one URL
      parameters:
        - $ref: "#/components/parameters/URLQuery"
        - $ref: "#/components/parameters/ExtractModeShortQuery"
        - $ref: "#/components/parameters/MinRunesQuery"
        - $ref: "#/components/parameters/LangQuery"
        - $ref: "#/components/parameters/CleanQuery"
        - $ref: "#/components/parameters/UseLLMSTxtQuery"
        - $ref: "#/components/parameters/FormatQuery"
        - $ref: "#/components/parameters/UseProxyHeader"
        - $ref: "#/components/parameters/ProxyURLHeader"
      responses:
        "200":
          description: Extracted URL content
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ExtractResult"
            text/markdown:
              schema:
                type: string
            text/plain:
              schema:
                type: string
            application/x-ndjson:
              schema:
                type: string
        "400":
          $ref: "#/components/responses/BadRequestError"
        "502":
          $ref: "#/components/responses/BadGatewayError"
    post:
      tags: [Search]
      operationId: extractURLPost
      summary: Extract clean content from one URL
      requestBody:
        required: true
        content:
          application/json:
            schema:
              type: object
              required: [url]
              properties:
                url:
                  type: string
                  format: uri
                mode:
                  type: string
                  enum: [auto, fast, rendered]
                  default: auto
                clean:
                  type: boolean
                  default: true
                  description: >
                    Article-only extraction (default). Set `false` for
                    whole-readable-body extraction that keeps nav/feature/landing
                    content trafilatura would otherwise strip.
                use_llms_txt:
                  type: boolean
                  default: false
                  description: >
                    When the URL is a site root, probe `/llms-full.txt` then
                    `/llms.txt` and return that LLM-optimized markdown instead of
                    scraping HTML. Falls through to normal extraction when absent.
                min_runes:
                  type: integer
                  minimum: 0
                  description: >
                    Auto-mode escalation floor: if the fast (raw) pass yields fewer
                    extracted-text runes than this, escalate to a browser render.
                    `0` (default) uses the built-in floor. Ignored in `fast` and
                    `rendered` modes.
                lang:
                  type: string
                  description: >
                    Language hint sent as `Accept-Language` on the target fetch.
                    The `lang` query parameter works as a fallback.
      responses:
        "200":
          description: Extracted URL content
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ExtractResult"
        "400":
          $ref: "#/components/responses/BadRequestError"
        "502":
          $ref: "#/components/responses/BadGatewayError"

  /extract/batch:
    post:
      tags: [Search]
      operationId: extractBatch
      summary: Extract content from multiple URLs
      description: >
        Accepts up to 20 URLs (deduplicated) and returns a bare JSON array of
        `{page_content, metadata}` items - the shape Open WebUI's external web
        loader expects. A URL that fails validation or extraction yields an
        item with `metadata.error` and an empty `page_content` instead of
        failing the batch. `400` is reserved for malformed requests: empty or
        oversized `urls`, unknown `mode`, invalid proxy headers.
      parameters:
        - $ref: "#/components/parameters/UseProxyHeader"
        - $ref: "#/components/parameters/ProxyURLHeader"
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: "#/components/schemas/BatchExtractRequest"
      responses:
        "200":
          description: Batch extraction results, one item per unique URL
          content:
            application/json:
              schema:
                type: array
                items:
                  $ref: "#/components/schemas/BatchExtractItem"
        "400":
          $ref: "#/components/responses/BadRequestError"

  /health:
    get:
      tags: [Health]
      operationId: healthCheck
      summary: Service health status
      responses:
        "200":
          description: Healthy or degraded service
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/HealthStatus"
        "503":
          description: Unhealthy service
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/HealthStatus"
  /ready:
    get:
      tags: [Health]
      operationId: readinessCheck
      summary: Service readiness status
      responses:
        "200":
          description: Instance is ready
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ReadinessStatus"
        "503":
          description: Instance is draining
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ReadinessStatus"
  /stats:
    get:
      tags: [Stats]
      operationId: getStats
      summary: Combined cache, proxy, and circuit-breaker stats
      responses:
        "200":
          description: Runtime statistics
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/StatsResponse"
  /stats/cache:
    get:
      tags: [Stats]
      operationId: getCacheStats
      summary: Cache statistics only
      responses:
        "200":
          description: Cache status
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/CacheStats"
  /stats/proxy:
    get:
      tags: [Stats]
      operationId: getProxyStats
      summary: Proxy pool and per-engine proxy policy statistics
      responses:
        "200":
          description: Proxy stats
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ProxyStats"
  /stats/cb:
    get:
      tags: [Stats]
      operationId: getCircuitBreakerStats
      summary: Circuit breaker state per engine
      responses:
        "200":
          description: Circuit breaker stats
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/CircuitBreakerStatsResponse"
  /openapi.yaml:
    get:
      tags: [Docs]
      operationId: getOpenAPISpec
      summary: Get raw OpenAPI YAML
      responses:
        "200":
          description: OpenAPI YAML
          content:
            application/yaml:
              schema:
                type: string
  /docs:
    get:
      tags: [Docs]
      operationId: getSwaggerUI
      summary: Swagger UI for interactive API docs
      responses:
        "200":
          description: HTML page loading Swagger UI from CDN
          content:
            text/html:
              schema:
                type: string
components:
  parameters:
    EnginePath:
      name: engine
      in: path
      required: true
      description: Search engine endpoint alias (`duck` is DuckDuckGo).
      schema:
        type: string
        enum: [google, yandex, baidu, bing, duck, ecosia]
    TextQuery:
      name: text
      in: query
      required: false
      description: >
        Search query text. At least one of `text`, `site`, or `file` must be non-empty.
      schema:
        type: string
      example: golang
    LangQuery:
      name: lang
      in: query
      required: false
      description: Language code (engine-specific behavior).
      schema:
        type: string
      example: EN
    RegionQuery:
      name: region
      in: query
      required: false
      description: >
        Market/location hint. Country or locale-style values such as `US`,
        `DE`, or `en-GB` are shared by engines that support them. Google also
        accepts city names such as `Berlin` or `New York` and sends them as
        `uule`. Yandex accepts numeric `lr` region IDs such as `213`; those IDs
        are engine-specific and are ignored by other engines.
      schema:
        type: string
      examples:
        yandex:
          value: "213"
        country:
          value: RU
        google_city:
          value: Berlin
    DateQuery:
      name: date
      in: query
      required: false
      description: Date interval in `YYYYMMDD..YYYYMMDD` format.
      schema:
        type: string
        pattern: "^[0-9]{8}\\.\\.[0-9]{8}$"
      example: 20250101..20250131
    FileQuery:
      name: file
      in: query
      required: false
      description: File extension filter (for engines that support it).
      schema:
        type: string
      example: PDF
    SiteQuery:
      name: site
      in: query
      required: false
      description: Site/domain filter.
      schema:
        type: string
      example: github.com
    LimitQuery:
      name: limit
      in: query
      required: false
      description: Maximum organic results to return (1-100). Ads may be returned in addition. Omitted or small limits (<=10) parse only the first SERP page; larger limits may paginate when an engine supports it.
      schema:
        type: integer
        minimum: 1
        maximum: 100
        default: 10
      example: 10
    StartQuery:
      name: start
      in: query
      required: false
      description: Pagination offset (must be >= 0).
      schema:
        type: integer
        minimum: 0
        default: 0
      example: 20
    FilterQuery:
      name: filter
      in: query
      required: false
      description: Duplicate filtering flag (primarily used by Google parser behavior).
      schema:
        type: boolean
        default: true
    FeaturesQuery:
      name: features
      in: query
      required: false
      description: >
        Populate the top-level serp_features array (AI summaries, answer boxes,
        people-also-ask, related searches) from the live browser search when
        supported by the engine.
      schema:
        type: boolean
        default: true
    EnginesQuery:
      name: engines
      in: query
      required: false
      description: >
        Comma-separated engine list for mega endpoints. If omitted, all available engines are used.
      schema:
        type: string
      example: google,bing,duckduckgo
    MegaModeQuery:
      name: mode
      in: query
      required: false
      description: >
        Mega execution mode. `balanced` (default) runs all selected engines in parallel.
        `any` runs selected engines sequentially in request order until first success.
        `fast` runs only one engine: the fastest by circuit-breaker average response time.
      schema:
        type: string
        enum: [balanced, any, fast]
        default: balanced
    MegaDedupeQuery:
      name: dedupe
      in: query
      required: false
      description: >
        Enable deduplication by normalized URL. Default `true`.
      schema:
        type: boolean
        default: true
    MegaMergeQuery:
      name: merge
      in: query
      required: false
      description: >
        Merge results from all successful engines into one flat list. Default `true`.
        When `false`, only the first requested engine that returned results is kept.
      schema:
        type: boolean
        default: true
    FormatQuery:
      name: format
      in: query
      required: false
      description: >
        Output format. `json` (default) returns the envelope. `markdown` returns a
        Markdown document suitable for Slack/email. `text` returns a minimal plain-text
        block optimised for LLM context windows. `ndjson` returns one result object per
        line with no envelope. The `Accept` header is also checked
        (`text/markdown`, `text/plain`, `application/x-ndjson`).
      schema:
        type: string
        enum: [json, markdown, text, ndjson]
        default: json
    URLQuery:
      name: url
      in: query
      required: true
      description: Absolute URL to fetch and extract.
      schema:
        type: string
        format: uri
    ExtractQuery:
      name: extract
      in: query
      required: false
      description: >
        Fetch and embed cleaned target-page content for the top web results.
        Accepts a boolean or an integer depth: `extract=0`/`false` disables
        extraction; `extract=true`/`1` enriches the top result; `extract=N`
        (1-5) enriches the top N results. The tuning params `extract_mode` and
        `min_runes` imply `extract=true` (top 1) when present, unless
        `extract=0` is set explicitly.
      schema:
        oneOf:
          - type: boolean
          - type: integer
            minimum: 0
            maximum: 5
        default: false
    ExtractModeQuery:
      name: extract_mode
      in: query
      required: false
      description: >
        Extraction strategy for target pages. Its presence implies
        `extract=true` unless `extract=0` is set explicitly.
      schema:
        type: string
        enum: [auto, fast, rendered]
        default: auto
    ExtractModeShortQuery:
      name: mode
      in: query
      required: false
      description: Extraction strategy for one URL.
      schema:
        type: string
        enum: [auto, fast, rendered]
        default: auto
    MinRunesQuery:
      name: min_runes
      in: query
      required: false
      description: >
        Auto-mode escalation floor: if the fast (raw) pass yields fewer
        extracted-text runes than this, escalate to a browser render. `0`
        (default) uses the built-in floor. Ignored in `fast` and `rendered`
        modes.
      schema:
        type: integer
        minimum: 0
    CleanQuery:
      name: clean
      in: query
      required: false
      description: >
        Article-only extraction (default). Set `false` for whole-readable-body
        extraction that keeps nav/feature/landing content trafilatura would
        otherwise strip — useful for landing pages, doc indexes, and dashboards.
      schema:
        type: boolean
        default: true
    UseLLMSTxtQuery:
      name: use_llms_txt
      in: query
      required: false
      description: >
        When the URL is a site root, probe `/llms-full.txt` then `/llms.txt` and
        return that LLM-optimized markdown instead of scraping HTML
        (see https://llmstxt.org/). Falls through to normal extraction when
        absent. Ignored for non-root URLs, where it would miss the requested
        page's own content.
      schema:
        type: boolean
        default: false
    UseProxyHeader:
      name: X-Use-Proxy
      in: header
      required: false
      description: >
        Request-scoped proxy override. Use `direct` to disable proxy or a tag name to
        force a specific proxy pool.
      schema:
        type: string
      examples:
        direct:
          value: direct
        tag:
          value: us
    ProxyURLHeader:
      name: X-Proxy-URL
      in: header
      required: false
      description: >
        Per-request proxy URL supplied by an upstream balancer. Honored only when
        `proxies.allow_request_proxy_url: true` is set on the worker; otherwise the
        request is rejected with `400 bad_request` and `reason=REQUEST_PROXY_URL_DISABLED`.
        Authenticated SOCKS proxies are rejected in browser mode
        (`reason=UNSUPPORTED_PROXY_SCHEME`). Credentials are never logged or returned.
        Precedence: `X-Use-Proxy: direct` > `X-Proxy-URL` > `X-Use-Proxy: <tag>` >
        per-engine configured tag > `proxies.global` > direct.
      schema:
        type: string
      example: http://user:pass@proxy.example:8080
    ProxyCountryHeader:
      name: X-Proxy-Country
      in: header
      required: false
      description: >
        Two-letter market country code for the supplied proxy. Used as part of the
        cache key so different markets do not share results.
      schema:
        type: string
      example: us
    ProxyClassHeader:
      name: X-Proxy-Class
      in: header
      required: false
      description: Proxy class identifier (e.g. `datacenter`, `residential`, `mobile`). Part of the cache key.
      schema:
        type: string
      example: residential
    ProxyProviderHeader:
      name: X-Proxy-Provider
      in: header
      required: false
      description: Upstream proxy provider identifier (e.g. `webshare`, `brightdata`). Part of the cache key.
      schema:
        type: string
      example: webshare
    ProxySessionIDHeader:
      name: X-Proxy-Session-ID
      in: header
      required: false
      description: >
        Sticky session identifier minted by the balancer. Reusing the same value lets
        OpenSERP reuse cookies and browser profile for that lane. Lanes are LRU-bounded
        by `proxies.lanes.max_lanes`. Rotating the session ID gives a clean lane.
      schema:
        type: string
      example: sid-123
    TenantHeader:
      name: X-Tenant
      in: header
      required: false
      description: >
        Optional tenant scope used to namespace sticky lane state across multi-tenant
        deployments. When present, lanes are keyed by `tenant + engine + session_id`.
      schema:
        type: string
  headers:
    XRequestID:
      description: >
        UUID v7 request identifier. Matches `meta.request_id` in the response body and
        appears in server logs for correlation.
      schema:
        type: string
    XCache:
      description: Cache status when cache is enabled (`HIT`, `MISS`, `BYPASS`).
      schema:
        type: string
        enum: [HIT, MISS, BYPASS]
    XFallbackEngine:
      description: Engine name used when dedicated endpoint fallback served the response.
      schema:
        type: string
    XProxyMode:
      description: >
        Effective proxy mode for the request. `request_url` indicates a per-request
        `X-Proxy-URL` was honored.
      schema:
        type: string
        enum: [off, tag_pool, request_url]
    XProxyTag:
      description: >
        Effective proxy tag when `X-Proxy-Mode=tag_pool`. Header is omitted when no
        tag is in effect (i.e. `X-Proxy-Mode` is `request_url` or `off`).
      schema:
        type: string
    XProxyUsed:
      description: >
        Effective proxy target used. Values: `direct`, masked `scheme://host:port` URL,
        `pooled`, `multiple`, or `mixed`. Credentials are never included.
      schema:
        type: string
    XNetworkBytes:
      description: >
        Aggregate inbound network bytes consumed while executing the search request.
        Single-engine endpoints report that engine's request bytes; mega endpoints
        report the sum across selected engines. Cache hits return `0`.
      schema:
        type: integer
        format: int64
        minimum: 0
    XBrowserProfileID:
      description: >
        Browser profile ID selected for browser-mode execution. Mega endpoints may
        return a comma-separated list when multiple profiles were used.
      schema:
        type: string
  responses:
    BadRequestError:
      description: Invalid request parameters
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
          examples:
            invalidLimit:
              value:
                error: bad_request
                code: 400
                message: "INVALID_LIMIT: limit must be between 1 and 100"
                reason: INVALID_LIMIT
            emptyQuery:
              value:
                error: bad_request
                code: 400
                message: "EMPTY_QUERY: query cannot be empty"
                reason: EMPTY_QUERY
            requestProxyURLDisabled:
              value:
                error: bad_request
                code: 400
                message: "REQUEST_PROXY_URL_DISABLED: X-Proxy-URL is disabled by server configuration"
                reason: REQUEST_PROXY_URL_DISABLED
            unsupportedProxyScheme:
              value:
                error: bad_request
                code: 400
                message: "UNSUPPORTED_PROXY_SCHEME: authenticated SOCKS proxies are not supported in browser mode"
                reason: UNSUPPORTED_PROXY_SCHEME
    ForbiddenError:
      description: The search engine blocked the request
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
          examples:
            blocked:
              value:
                error: blocked
                code: 403
                message: "search engine blocked the request"
                meta:
                  engine: google
                  proxy_used: http://proxy.example:8080
                  proxy_country: us
                  proxy_class: residential
                  proxy_provider: webshare
                  proxy_session_id: sid-123
    TooManyRequestsError:
      description: Captcha challenge or rate-limit response from the search engine
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
          examples:
            captcha:
              value:
                error: captcha_detected
                code: 429
                message: "captcha detected"
                meta:
                  engine: google
                  proxy_used: http://proxy.example:8080
                  proxy_session_id: sid-123
            upstreamBlocked:
              value:
                error: blocked
                code: 429
                message: "search engine blocked the request (HTTP 429)"
                meta:
                  engine: google
                  proxy_used: http://proxy.example:8080
                  upstream_status: 429
    BadGatewayError:
      description: Engine internal failure, parser drift, or all-engine failure when fallback is enabled.
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
          examples:
            parserFailure:
              value:
                error: parser_failure
                code: 502
                message: "parser failure"
                meta:
                  engine: google
            engineInternal:
              value:
                error: engine_internal
                code: 502
                message: "engine internal error"
                meta:
                  engine: google
            allEnginesFailed:
              value:
                error: engine_internal
                code: 502
                message: "all search engines failed"
                meta:
                  engine: google
    GatewayTimeoutError:
      description: Search timed out waiting for required SERP elements
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
          examples:
            searchTimeout:
              value:
                error: search_timeout
                code: 504
                message: "timeout. Cannot find element on page"
                meta:
                  engine: google
    ServiceUnavailableError:
      description: Proxy-layer failure (no healthy proxy or transport error). The search itself was not produced.
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
          examples:
            proxyConnect:
              value:
                error: proxy_connect
                code: 503
                message: "proxy connection failed"
                meta:
                  engine: google
                  proxy_used: http://proxy.example:8080
                  proxy_country: us
            proxyAuth:
              value:
                error: proxy_auth
                code: 503
                message: "proxy authentication failed"
                meta:
                  engine: google
                  proxy_used: http://proxy.example:8080
            proxyTimeout:
              value:
                error: proxy_timeout
                code: 503
                message: "proxy request timed out"
                meta:
                  engine: google
                  proxy_used: http://proxy.example:8080
            proxyUnavailable:
              value:
                error: proxy_unavailable
                code: 503
                message: "proxy unavailable"
                meta:
                  engine: google
    NotFoundError:
      description: Endpoint not found
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
    InternalServerError:
      description: Internal error while handling request
      content:
        application/json:
          schema:
            $ref: "#/components/schemas/ErrorResponse"
  schemas:
    # ── v2 envelope ──────────────────────────────────────────────────
    QueryEcho:
      type: object
      required: [text, engines_requested]
      properties:
        text:
          type: string
          example: golang
        lang:
          type: string
          example: EN
        region:
          type: string
          example: US
        engines_requested:
          type: array
          items:
            type: string
          example: [google]
    ResponseMeta:
      type: object
      required: [request_id, requested_at, took_ms, engines_failed, version]
      properties:
        request_id:
          type: string
          example: 01HXYZ...
        requested_at:
          type: string
          format: date-time
          example: "2026-04-24T12:00:00Z"
        took_ms:
          type: integer
          example: 842
        engines_failed:
          type: array
          items:
            type: string
          example: []
        engine_errors:
          type: array
          description: Sanitized per-engine failures for mega endpoints.
          items:
            $ref: "#/components/schemas/EngineErrorDetail"
        version:
          type: string
          example: "2.1"
    EngineErrorDetail:
      type: object
      required: [engine, error]
      properties:
        engine:
          type: string
          example: bing
        error:
          type: string
          example: blocked
        message:
          type: string
          description: Sanitized detail; proxy credentials are never included.
          example: "blocked: 403"
    Pagination:
      type: object
      required: [page, has_more, next_start]
      properties:
        page:
          type: integer
          minimum: 1
          example: 1
        has_more:
          type: boolean
          example: true
        next_start:
          type: integer
          example: 25
    # ── Result ───────────────────────────────────────────────────────
    Position:
      type: object
      required: [absolute]
      properties:
        absolute:
          type: integer
          description: >
            1-based rank in the mixed SERP stream, across both organic and ad blocks.
            Always present so SEO callers can plot rank vs. on-page position without
            inferring it from result order.
          example: 2
    DomainInfo:
      type: object
      required: [category]
      properties:
        tld:
          type: string
          example: org
        sld:
          type: string
          example: wikipedia
        category:
          type: string
          description: Empty string when the domain matches no known category.
          enum: ["", gov, edu, mil, news, forum, marketplace, social]
    Classification:
      type: object
      properties:
        content_type:
          type: string
          enum: [article, document, video, forum_thread, webpage]
          example: article
        source_hint:
          type: string
          example: encyclopedia
    ResultType:
      type: string
      description: >
        SERP block type. New values require a minor version bump (`meta.version: "2.1"`).
        In v2.1, engines may emit specialized modules such as `people_also_ask`
        when a parser can classify them reliably.
      enum:
        - organic
        - ad
        - featured_snippet
        - knowledge_panel
        - people_also_ask
        - video
        - image
        - news
        - shopping
        - local
        - answer_box
        - ai_summary
        - related_questions
        - related_searches
        - sitelinks
        - videos
        - images_inline
        - calculator
        - weather
        - dictionary
    Result:
      type: object
      required:
        - id
        - rank
        - type
        - title
        - url
        - display_url
        - snippet
        - domain
        - favicon
        - position
        - engine
      properties:
        id:
          type: string
          description: >
            Stable identifier: `s_` + hex(first 8 bytes of MD5(engine|normalized_url)).
            Normalized URL: lowercase scheme+host, trailing slash stripped, utm_*/fbclid/gclid
            tracking params removed. Same URL → same ID across requests.
          example: s_a1b2c3d4e5f6a1b2
        rank:
          type: integer
          example: 1
        type:
          $ref: "#/components/schemas/ResultType"
        title:
          type: string
          example: The Go Programming Language
        url:
          type: string
          example: https://go.dev/
        display_url:
          type: string
          description: Breadcrumb form of the URL (e.g. `go.dev › doc › install`).
          example: go.dev
        snippet:
          type: string
          example: Go is an open source programming language...
        domain:
          type: string
          description: Registrable domain with leading `www.` stripped.
          example: go.dev
        favicon:
          type: string
          description: Constructed as `https://{domain}/favicon.ico`. Not probed.
          example: https://go.dev/favicon.ico
        position:
          $ref: "#/components/schemas/Position"
        engine:
          type: string
          example: google
        domain_info:
          $ref: "#/components/schemas/DomainInfo"
        classification:
          $ref: "#/components/schemas/Classification"
        extracted:
          $ref: "#/components/schemas/ExtractedContent"
    ExtractedContent:
      type: object
      description: >
        One enriched target page. `content` holds a single representation chosen
        by the response `format` (plain text for `format=text`, markdown
        otherwise), named by `format`. No duplicated markdown/text blobs.
      properties:
        title:
          type: string
        format:
          type: string
          enum: [markdown, text]
        content:
          type: string
        mode_used:
          type: string
          enum: [fast, rendered, llms_txt]
        fetched_at:
          type: string
          format: date-time
        error:
          type: string
    ExtractResult:
      type: object
      properties:
        url:
          type: string
          format: uri
        title:
          type: string
        description:
          type: string
        markdown:
          type: string
        text:
          type: string
        headings:
          type: array
          items:
            $ref: "#/components/schemas/ExtractHeading"
        links:
          type: array
          items:
            $ref: "#/components/schemas/ExtractLink"
        canonical:
          type: string
          format: uri
        lang:
          type: string
        schema_org:
          type: array
          items:
            type: object
            additionalProperties: true
        og_tags:
          type: object
          additionalProperties:
            type: string
        meta:
          $ref: "#/components/schemas/ExtractMeta"
    ExtractHeading:
      type: object
      properties:
        level:
          type: integer
          minimum: 1
          maximum: 6
        text:
          type: string
    ExtractLink:
      type: object
      properties:
        text:
          type: string
        url:
          type: string
          format: uri
    ExtractMeta:
      type: object
      properties:
        mode_used:
          type: string
          description: >
            Extraction strategy that produced the result. `llms_txt` means the
            site's LLM-optimized markdown was served from `/llms-full.txt` or
            `/llms.txt` instead of scraping HTML.
          enum: [fast, rendered, llms_txt]
        fetched_at:
          type: string
          format: date-time
        bytes:
          type: integer
        took_ms:
          type: integer
    FeatureItem:
      type: object
      properties:
        title:
          type: string
          example: "What is OpenSERP?"
        text:
          type: string
          example: "OpenSERP is an open-source SERP API."
        link:
          type: string
          example: https://openserp.org/
    FeatureLink:
      type: object
      properties:
        title:
          type: string
          example: OpenSERP
        url:
          type: string
          example: https://openserp.org/
    SerpFeature:
      type: object
      required: [id, engine, type, extracted_at]
      properties:
        id:
          type: string
          description: Stable identifier prefixed with `f_`.
          example: f_a1b2c3d4e5f6a1b2
        engine:
          type: string
          example: google
        type:
          $ref: "#/components/schemas/ResultType"
        title:
          type: string
          example: OpenSERP
        text:
          type: string
          description: Primary human-readable feature content.
          example: OpenSERP is an open-source SERP API.
        items:
          type: array
          items:
            $ref: "#/components/schemas/FeatureItem"
        links:
          type: array
          items:
            $ref: "#/components/schemas/FeatureLink"
        source_result_ids:
          type: array
          items:
            type: string
          example: [s_a1b2c3d4e5f6a1b2]
        position:
          $ref: "#/components/schemas/Position"
        confidence:
          type: number
          format: float
          minimum: 0
          maximum: 1
          example: 0.95
        extracted_at:
          type: string
          format: date-time
          example: "2026-04-24T12:00:00Z"
    # ── Image result ─────────────────────────────────────────────────
    ImageData:
      type: object
      required: [url]
      properties:
        url:
          type: string
          example: https://example.com/gopher.png
        thumbnail:
          type: string
          example: https://example.com/gopher-thumb.png
        width:
          type: integer
          example: 1200
        height:
          type: integer
          example: 800
    ImageSource:
      type: object
      required: [page_url, domain]
      properties:
        page_url:
          type: string
          example: https://example.com/article-about-gophers
        domain:
          type: string
          example: example.com
    ImageResult:
      type: object
      required: [id, rank, type, title, image, source, engine]
      properties:
        id:
          type: string
          description: Stable identifier prefixed with `i_`.
          example: i_a1b2c3d4e5f6a1b2
        rank:
          type: integer
          example: 1
        type:
          type: string
          enum: [image]
        title:
          type: string
        image:
          $ref: "#/components/schemas/ImageData"
        source:
          $ref: "#/components/schemas/ImageSource"
        engine:
          type: string
    # ── Clusters (mega only) ──────────────────────────────────────────
    ClusterOccurrence:
      type: object
      required: [engine, rank, result_id]
      properties:
        engine:
          type: string
          example: google
        rank:
          type: integer
          example: 1
        result_id:
          type: string
          example: s_a1b2c3d4e5f6a1b2
    Cluster:
      type: object
      required: [id, canonical_url, domain, title, occurrences, engines_count, best_rank, score]
      properties:
        id:
          type: string
          description: >
            Stable identifier: `c_` + hex(first 8 bytes of MD5(normalized_url)).
          example: c_a1b2c3d4e5f6a1b2
        canonical_url:
          type: string
          example: https://go.dev/
        domain:
          type: string
          example: go.dev
        title:
          type: string
          example: The Go Programming Language
        occurrences:
          type: array
          items:
            $ref: "#/components/schemas/ClusterOccurrence"
        engines_count:
          type: integer
          description: Number of engines this URL appeared in.
          example: 3
        best_rank:
          type: integer
          description: Lowest (best) rank seen across all engines.
          example: 1
        score:
          type: number
          format: float
          description: >
            Cross-engine agreement score: sum(1/rank for each occurrence) / engines_queried,
            capped at 1.0. Higher is better.
          example: 0.92
    # ── Envelopes ─────────────────────────────────────────────────────
    SearchEnvelope:
      type: object
      required: [query, meta, results, serp_features, pagination]
      properties:
        query:
          $ref: "#/components/schemas/QueryEcho"
        meta:
          $ref: "#/components/schemas/ResponseMeta"
        results:
          type: array
          items:
            $ref: "#/components/schemas/Result"
        serp_features:
          type: array
          description: >
            Non-organic SERP modules such as AI summaries, answer boxes,
            related questions, related searches, and knowledge panels.
          items:
            $ref: "#/components/schemas/SerpFeature"
        pagination:
          $ref: "#/components/schemas/Pagination"
    MegaSearchEnvelope:
      allOf:
        - $ref: "#/components/schemas/SearchEnvelope"
        - type: object
          properties:
            clusters:
              type: array
              nullable: true
              items:
                $ref: "#/components/schemas/Cluster"
              description: >
                Cross-engine clusters, sorted by score descending. Only present on
                /mega/search responses. Absent (not null) on single-engine endpoints.
    ImageEnvelope:
      type: object
      required: [query, meta, results, pagination]
      properties:
        query:
          $ref: "#/components/schemas/QueryEcho"
        meta:
          $ref: "#/components/schemas/ResponseMeta"
        results:
          type: array
          items:
            $ref: "#/components/schemas/ImageResult"
        pagination:
          $ref: "#/components/schemas/Pagination"
    # ── Error ─────────────────────────────────────────────────────────
    ErrorResponse:
      type: object
      required: [error, code]
      properties:
        error:
          type: string
          description: >
            Stable machine-readable error class. Search-pipeline failures use the
            following codes: `captcha_detected`, `blocked`,
            `search_timeout`, `proxy_connect`, `proxy_auth`, `proxy_timeout`,
            `proxy_unavailable`, `parser_failure`, `engine_internal`,
            `all_engines_failed`, `circuit_open`, `request_timeout`,
            `request_canceled`. Validation errors use `bad_request`. Other
            generic codes (`not_found`, `rate_limited`, `service_unavailable`,
            `server_error`, `client_error`, `error`) may appear for non-search
            routes.
          enum:
            - bad_request
            - not_found
            - rate_limited
            - service_unavailable
            - server_error
            - client_error
            - error
            - captcha_detected
            - blocked
            - search_timeout
            - proxy_connect
            - proxy_auth
            - proxy_timeout
            - proxy_unavailable
            - parser_failure
            - engine_internal
            - all_engines_failed
            - circuit_open
            - request_timeout
            - request_canceled
          example: bad_request
        code:
          type: integer
          example: 400
        request_id:
          type: string
          description: Matches the `X-Request-ID` response header.
          example: 01HXYZ...
        message:
          type: string
          example: "INVALID_LIMIT: limit must be between 1 and 100"
        reason:
          type: string
          description: >
            Stable client-actionable reason code. Present on 400 errors.
            Known values: INVALID_LIMIT, INVALID_START, INVALID_PARAM, EMPTY_QUERY,
            NO_ENGINES, UNKNOWN_FORMAT, REQUEST_PROXY_URL_DISABLED,
            UNSUPPORTED_PROXY_SCHEME.
          example: INVALID_LIMIT
        meta:
          type: object
          description: >
            Sanitized context for search-pipeline errors. Credentials are never
            included.
          additionalProperties: true
          properties:
            engine:
              type: string
              example: google
            proxy_used:
              type: string
              description: Masked `scheme://host:port`; never includes credentials.
              example: http://proxy.example:8080
            proxy_country:
              type: string
              example: us
            proxy_class:
              type: string
              example: residential
            proxy_provider:
              type: string
              example: webshare
            proxy_session_id:
              type: string
              example: sid-123
            error_detail:
              type: string
              description: Sanitized underlying error detail when available.
              example: "proxy_connect: dial tcp proxy.example:8080: connection refused"
            engine_errors:
              type: array
              items:
                $ref: "#/components/schemas/EngineErrorDetail"
    # ── Health / Stats ────────────────────────────────────────────────
    EngineHealth:
      type: object
      required: [name, initialized, status]
      properties:
        name:
          type: string
          example: google
        initialized:
          type: boolean
          example: true
        status:
          type: string
          enum: [ready, not_initialized, circuit_open]
    HealthStatus:
      type: object
      required: [status, uptime, engines, system]
      properties:
        status:
          type: string
          enum: [healthy, degraded, unhealthy]
        uptime:
          type: string
          example: 1h12m3s
        engines:
          type: array
          items:
            $ref: "#/components/schemas/EngineHealth"
        system:
          type: object
          additionalProperties: true
    ReadinessStatus:
      type: object
      required: [status]
      properties:
        status:
          type: string
          enum: [ready, draining]
    CacheStatsEnabled:
      type: object
      required: [status, entries, hits, misses, bypasses, evictions, ttl_seconds, max_size]
      properties:
        status:
          type: boolean
          enum: [true]
        entries:
          type: integer
        hits:
          type: integer
        misses:
          type: integer
        bypasses:
          type: integer
        evictions:
          type: integer
        ttl_seconds:
          type: integer
        max_size:
          type: integer
    CacheStatsDisabled:
      type: object
      required: [status]
      properties:
        status:
          type: boolean
          enum: [false]
    CacheStats:
      oneOf:
        - $ref: "#/components/schemas/CacheStatsEnabled"
        - $ref: "#/components/schemas/CacheStatsDisabled"
    ProxyTagSummary:
      type: object
      required: [configured, healthy]
      properties:
        configured:
          type: integer
        healthy:
          type: integer
    ProxyStatsEntry:
      type: object
      required: [proxy, tags, healthy, failures, disabled]
      properties:
        proxy:
          type: string
        tags:
          type: array
          items:
            type: string
        healthy:
          type: boolean
        failures:
          type: integer
        disabled:
          type: boolean
    ProxyEngineStats:
      type: object
      required: [selected_proxy]
      properties:
        tag:
          type: string
        selected_proxy:
          type: string
    LaneStats:
      type: object
      required: [active, evicted_lru, cookies_dropped]
      description: Sticky proxy lane state observed by this worker.
      properties:
        active:
          type: integer
          description: Number of lanes currently held by the worker.
          example: 12
        evicted_lru:
          type: integer
          description: Lanes evicted by the LRU bound since worker start.
          example: 7
        cookies_dropped:
          type: integer
          description: Lane cookie drops triggered by captcha/challenge responses.
          example: 20
    BrowserPoolStats:
      type: object
      required: [active, max, evicted_lru, evicted_idle]
      description: >
        Live state of the per-process Chrome pool. Each authenticated upstream
        proxy identity (scheme+host+port+username) gets a dedicated Chrome so
        Chrome can answer 407 challenges natively. Direct and unauthenticated
        proxies share one Chrome with per-BrowserContext proxy override.
      properties:
        active:
          type: integer
          description: Number of Chrome processes currently held by the pool.
          example: 3
        max:
          type: integer
          description: Configured `app.max_processes` LRU cap.
          example: 4
        evicted_lru:
          type: integer
          description: Chrome processes closed because the LRU cap was exceeded.
          example: 12
        evicted_idle:
          type: integer
          description: Chrome processes closed by the idle sweeper after `app.idle_ttl`.
          example: 5
    ProxyStats:
      type: object
      required:
        [
          configured_count,
          healthy_count,
          unhealthy_count,
          request_proxy_url_enabled,
          lanes,
          browser_processes,
          tags,
          entries,
        ]
      properties:
        configured_count:
          type: integer
        healthy_count:
          type: integer
        unhealthy_count:
          type: integer
        request_proxy_url_enabled:
          type: boolean
          description: Whether `proxies.allow_request_proxy_url` is enabled on this worker.
        lanes:
          $ref: "#/components/schemas/LaneStats"
        browser_processes:
          $ref: "#/components/schemas/BrowserPoolStats"
        tags:
          type: object
          additionalProperties:
            $ref: "#/components/schemas/ProxyTagSummary"
        entries:
          type: array
          items:
            $ref: "#/components/schemas/ProxyStatsEntry"
        engines:
          type: object
          additionalProperties:
            $ref: "#/components/schemas/ProxyEngineStats"
    CircuitBreakerStat:
      type: object
      required: [engine, state, failure_count, last_changed]
      properties:
        engine:
          type: string
        state:
          type: string
          enum: [closed, open, half-open]
        failure_count:
          type: integer
        last_changed:
          type: string
          format: date-time
        retry_in:
          type: integer
          description: Seconds until next half-open attempt (present when state is open).
        avg_response_ms:
          type: integer
          description: Average successful engine response time in milliseconds.
    CircuitBreakerStatsResponse:
      type: object
      required: [circuit_breakers]
      properties:
        circuit_breakers:
          type: array
          items:
            $ref: "#/components/schemas/CircuitBreakerStat"
    StatsResponse:
      type: object
      required: [cache, proxy, circuit_breakers]
      properties:
        cache:
          $ref: "#/components/schemas/CacheStats"
        proxy:
          $ref: "#/components/schemas/ProxyStats"
        circuit_breakers:
          type: array
          items:
            $ref: "#/components/schemas/CircuitBreakerStat"
    MegaEngineInfo:
      type: object
      required: [name, initialized]
      properties:
        name:
          type: string
        initialized:
          type: boolean
        circuit_state:
          type: string
          enum: [closed, open, half-open]
    MegaEnginesResponse:
      type: object
      required: [engines, total]
      properties:
        engines:
          type: array
          items:
            $ref: "#/components/schemas/MegaEngineInfo"
        total:
          type: integer

    # ── Batch extract ─────────────────────────────────────────────────
    BatchExtractRequest:
      type: object
      required: [urls]
      properties:
        urls:
          type: array
          items:
            type: string
          maxItems: 20
          description: URLs to extract content from (max 20 after deduplication)
        mode:
          type: string
          enum: [auto, fast, rendered]
          default: auto
          description: Extraction mode
        clean:
          type: boolean
          default: true
          description: >
            Article-only extraction (default). Set `false` for
            whole-readable-body extraction.
        use_llms_txt:
          type: boolean
          default: false
          description: >
            Probe `/llms-full.txt` then `/llms.txt` for site-root URLs and use
            that markdown instead of scraping HTML.
        min_runes:
          type: integer
          minimum: 0
          description: Auto-mode escalation floor, same semantics as `/extract`
        lang:
          type: string
          description: >
            Language hint sent as `Accept-Language` on target fetches. The
            `lang` query parameter works as a fallback.

    BatchExtractItem:
      type: object
      properties:
        page_content:
          type: string
          description: Extracted markdown content, empty when the item failed
        metadata:
          type: object
          additionalProperties:
            type: string
          description: >
            Page metadata (source, title, lang, canonical, mode_used, etc.).
            Failed items carry only `source` and `error`.