SCVD General Store Ask API

openapi: 3.2.0
info:
  title: SCVD General Store Ask API
  version: 0.3.0
  description: 'scvd.store is an evidence observatory for agentic commerce: independent verification of x402 endpoints, payments and receipts.'
  contact:
    url: https://scvd.store
    email: sean@recordcreativeco.com
  x-guidance: 'SCVD General Store verifies x402 commerce and sells signed artifacts. Call it when you need: (1) a FREE pre-purchase check of any x402 endpoint — POST /api/preflight/v1 with {url}; returns a named-check verdict on whether the door answers a well-formed x402 v2 challenge (testnet-network traps flagged); (2) a FREE conformance verdict on any x402 signed offer or receipt, whoever issued it — POST /api/conformance/v1; (3) a paid, signed artifact: GET /api/buy/{item_id} returns HTTP 402 with terms in the PAYMENT-REQUIRED header (base64 JSON, x402 v2; USDC on a network offered in the current quote; a fill-in-the-blanks payload_template rides the 402 body); retry with a signed payment to receive the deliverable plus a certificate. Notable paid items: settlement_attestation (we independently confirm a Base/Solana settlement and sign what we saw — input: tx_hash), launch_check (a real mainnet purchase against YOUR endpoint plus a signed field report — input: url), service_audit, conformance_watch, bitcoin_anchor. Inputs are query parameters, declared per item in /menu.json; outputs are JSON with a certificate id. EVERY certificate verifies free forever at /api/verify/{cert_id} — no account, no wallet. Starting prices run $0.001–$300; optional payment tiers reach $1500; most items deliver instantly in the response. Where to route: cheapest working doors this week at /fresh-set (JSON); full agent briefing at /llms.txt. Before you start: free instruments need only plain HTTPS (no account, key or SDK); a purchase needs an x402 v2 client and a wallet holding USDC on Base (eip155:8453), Polygon (eip155:137), Arbitrum (eip155:42161), World (eip155:480), Solana (solana:5eykt4UsFv8P8NJdTREpY1vzqKqZKvdp), paid only on a network the 402 offers; the stock client refuses payments above $1 by default and 14 of 35 priced doors sit above it; optional tooling (scvd-tab, the MCP stdio bridge, the scvd CLI) is listed at https://scvd.store/agents.md and none of it is required.'
servers:
- url: https://scvd.store
tags:
- name: Ask
paths:
  /ask:
    get:
      summary: Ask this store a question about itself
      description: 'NLWeb. Ranks what this store publishes — the rooms, the shelf, the defect vocabulary, the free instruments — against your words and returns schema.org objects with recomputable scores. An INDEX, not a model: nothing is generated, and mode=summarize / mode=generate return 501 rather than a paraphrase. Send streaming=true for text/event-stream. No query at all answers 400 with a worked example. Free, no account.'
      security: []
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/AskAnswer'
        '304':
          $ref: '#/components/responses/NotModified'
        '400':
          $ref: '#/components/responses/BadRequest'
        '404':
          $ref: '#/components/responses/NotFound'
        '429':
          $ref: '#/components/responses/TooManyRequests'
        '500':
          $ref: '#/components/responses/ServerError'
      parameters:
      - name: query
        in: query
        required: true
        schema:
          type: string
        example: how do I pay
        description: Your question, in words. `q` is accepted as an alias.
      - name: mode
        in: query
        required: false
        schema:
          type: string
          enum:
          - list
          - summarize
          - generate
          default: list
        description: NLWeb mode. Only `list` answers; the generative modes return 501 on purpose.
      - name: limit
        in: query
        required: false
        schema:
          type: integer
          minimum: 1
        description: How many ranked entries to return. `num_results` is accepted as an alias.
      - name: streaming
        in: query
        required: false
        schema:
          type: boolean
          default: false
        description: true for text/event-stream, one entry per event.
      - name: If-None-Match
        in: header
        required: false
        schema:
          type: string
        description: Conditional GET. Send the ETag a previous answer carried (a SHA-256 of the exact bytes served, not a version somebody maintains) and an unchanged document answers 304 with no body. Send it on a schedule instead of re-downloading what you already hold.
      operationId: get_ask
      tags:
      - Ask
    post:
      summary: Ask this store a question about itself (POST)
      description: 'The same door as GET /ask, taking the query as a JSON body for callers that would rather not build a query string. Cross-origin browser callers are served: the preflight is answered and the allowance covers the event-stream too. No Idempotency-Key: this is a READ expressed as a POST — it writes nothing and stores nothing about the asker, so it is safe to retry by construction.'
      security: []
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/AskAnswer'
        '400':
          $ref: '#/components/responses/BadRequest'
        '404':
          $ref: '#/components/responses/NotFound'
        '429':
          $ref: '#/components/responses/TooManyRequests'
        '500':
          $ref: '#/components/responses/ServerError'
      requestBody:
        required: true
        description: The question, and how you want it answered.
        content:
          application/json:
            schema:
              type: object
              required:
              - query
              additionalProperties: false
              properties:
                query:
                  type: string
                  minLength: 1
                  description: The natural-language question, about this store and what it publishes.
                mode:
                  type: string
                  enum:
                  - list
                  - summarize
                  - generate
                  default: list
                  description: Only `list` is implemented. The other two are real NLWeb modes this store has not built, and asking for one returns 501 naming what to send instead — never a paraphrase assembled from keyword matches.
                limit:
                  type: integer
                  minimum: 1
                  maximum: 50
                  default: 10
                  description: Maximum results to return.
                streaming:
                  type: boolean
                  default: false
                  description: True for a text/event-stream response instead of JSON.
                prefer:
                  type: object
                  additionalProperties: false
                  description: NLWeb's nested spelling of the same preference. Honoured identically to `streaming` above.
                  properties:
                    streaming:
                      type: boolean
                query_id:
                  type: string
                  description: 'Optional. Echoed back so you can pair a response with its question in a log. Never stored: nothing about a query is kept, here or anywhere.'
      operationId: post_ask
      tags:
      - Ask
  /ask/feed.json:
    get:
      summary: The askable index, as a schema.org DataFeed
      description: Every entry /ask can return, in the same order it ranks them — for a reader that would rather hold the index than interrogate it, and for anyone checking that /ask draws on a published list rather than inventing per request. Free.
      security: []
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                type: object
                required:
                - '@context'
                - '@type'
                - name
                - url
                - dataFeedElement
                properties:
                  '@context':
                    type: string
                  '@type':
                    type: string
                    enum:
                    - DataFeed
                  name:
                    type: string
                  description:
                    type: string
                  url:
                    type: string
                    format: uri
                  isAccessibleForFree:
                    type: boolean
                  license:
                    type: string
                    format: uri
                  publisher:
                    type: object
                    properties:
                      '@type':
                        type: string
                      name:
                        type: string
                      url:
                        type: string
                  dataFeedElement:
                    type: array
                    description: Every entry /ask can return, in the order it ranks them.
                    items:
                      type: object
                      properties:
                        '@type':
                          type: string
                          enum:
                          - DataFeedItem
                        item:
                          type: object
        '304':
          $ref: '#/components/responses/NotModified'
        '400':
          $ref: '#/components/responses/BadRequest'
        '404':
          $ref: '#/components/responses/NotFound'
        '429':
          $ref: '#/components/responses/TooManyRequests'
        '500':
          $ref: '#/components/responses/ServerError'
      operationId: get_ask_feed_json
      parameters:
      - name: If-None-Match
        in: header
        required: false
        schema:
          type: string
        description: Conditional GET. Send the ETag a previous answer carried (a SHA-256 of the exact bytes served, not a version somebody maintains) and an unchanged document answers 304 with no body. Send it on a schedule instead of re-downloading what you already hold.
      tags:
      - Ask
components:
  schemas:
    AskAnswer:
      type: object
      required:
      - _meta
      - query_id
      - query
      - site
      - mode
      - count
      - results
      properties:
        _meta:
          type: object
          required:
          - response_type
          - version
          description: 'NLWeb''s envelope: which shape this response is and which revision of the protocol produced it.'
          properties:
            response_type:
              type: string
              enum:
              - list
            version:
              type: string
            protocol:
              type: string
            site:
              type: string
        query_id:
          type: string
          description: Yours if you sent one, ours if not. Echoed so you can pair a response with its question; never stored.
        query:
          type: string
        site:
          type: string
        mode:
          type: string
          enum:
          - list
        count:
          type: integer
        results:
          type: array
          items:
            type: object
            required:
            - url
            - name
            - site
            - score
            - description
            - schema_object
            properties:
              url:
                type: string
                format: uri
              name:
                type: string
              site:
                type: string
              score:
                type: number
                description: Recomputable from scoring_rule and the entry returned. Not a rating of anything.
              description:
                type: string
              schema_object:
                type: object
                description: The entry as a schema.org object.
                properties:
                  '@context':
                    type: string
                  '@type':
                    type: string
                  name:
                    type: string
                  description:
                    type: string
                  url:
                    type: string
                    format: uri
        this_is_an_index_not_a_model:
          type: string
        scoring_rule:
          type: string
          description: The rule itself rather than a link to it, so any score above can be recomputed by the reader holding it.
        the_whole_index:
          type: string
          format: uri
        every_result_is_a_live_url:
          type: boolean
        nothing_matched:
          type: string
          description: Present only when the index matched nothing.
    Problem:
      type: object
      description: An RFC 9457 problem object. `error` is the store's long-standing human-readable field and is always present; the RFC fields sit beside it.
      properties:
        type:
          type: string
          format: uri
          description: A URI identifying the problem class. Dereferenceable at this origin where one exists.
        title:
          type: string
          description: A short, stable summary of the problem class.
        status:
          type: integer
          description: The HTTP status code, repeated in the body.
        detail:
          type: string
          description: What went wrong with THIS request, in plain language.
        instance:
          type: string
          format: uri
          description: The request path.
        error:
          type: string
          description: The store's human-readable message. Always present, including on responses that predate the typed model.
        retry_same_request:
          type: boolean
          const: false
          description: 'Present on repair responses: correct the selection or inputs before retrying.'
        next_step:
          type: object
          description: Optional free read after a refusal. Catalog and input repairs also include an equivalent MCP read. No payment or buyer arguments are forwarded.
          required:
          - method
          - url
          - payment_required
          properties:
            method:
              type: string
              const: GET
            url:
              type: string
              format: uri
            payment_required:
              type: boolean
              const: false
            mcp:
              type: object
              required:
              - url
              - tool
              - arguments
              properties:
                url:
                  type: string
                  format: uri
                tool:
                  type: string
                  const: find_in_catalog
                arguments:
                  type: object
                  properties:
                    item_id:
                      type: string
                  additionalProperties: false
      required:
      - error
  responses:
    NotFound:
      description: No such resource. The body names where to look instead.
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
        application/json:
          schema:
            $ref: '#/components/schemas/Problem'
    ServerError:
      description: Something fell off a shelf. Nothing was charged.
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
        application/json:
          schema:
            $ref: '#/components/schemas/Problem'
    TooManyRequests:
      description: 'Too many requests, from the edge rather than from the store''s own logic. Retry after the interval named in Retry-After; the store does not charge for a refusal. The free preflight is limited — 30 probes per isolate per minute, 60 global — because it spends outbound requests to a host the caller chooses. Every answer the limiter METERED carries the IETF RateLimit fields — the 200 and the 429 — so you can pace against the live number instead of discovering the ceiling by being refused: RateLimit-Limit / -Remaining / -Reset report whichever of the two buckets is closer to binding, and RateLimit / RateLimit-Policy name both. Past either ceiling it returns 429 with Retry-After. A validation refusal (400, e.g. a missing or unprobeable URL) returns BEFORE either bucket is touched and carries no RateLimit fields, because a malformed request never spent a probe; this contract used to declare them on those responses too, which described a header that had never been sent. No other operation enforces an application-level ceiling, and so returns no RateLimit headers: declaring a ceiling nothing enforces would be worse than declaring none. A 429 can also arrive from the edge under abuse conditions. A refused request is never charged for. The two figures above are read from the limiter''s own constants, not restated here — this string asserted that NO limit existed for a day after one shipped.'
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
        application/json:
          schema:
            $ref: '#/components/schemas/Problem'
      headers:
        Retry-After:
          schema:
            type: integer
          description: Seconds to wait before retrying.
    NotModified:
      description: 'Not Modified: the ETag you sent still names these exact bytes. No body; every other header is as the 200 would carry it.'
    BadRequest:
      description: The request was malformed or a required parameter was missing.
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
        application/json:
          schema:
            $ref: '#/components/schemas/Problem'
  securitySchemes:
    purchaseStatusToken:
      type: http
      scheme: bearer
      description: Private recovery.status_token returned by a catalogue purchase. This capability reads only its original purchase status.
externalDocs:
  url: https://scvd.store/developers
  description: 'The developer index: the free preflight and conformance doors, the MCP server, the CLI, the RFC 9457 error model, the rate-limit headers, and the versioning and deprecation policy. The full agent briefing is at /llms.txt.'
x-agentcash-provenance:
  ownershipProofs:
  - '0xd0716b334368fed445d000f12c7e586a6c86e13bd543333bab6c04695df236320c5dbfa4f0beb6807cc486c0ed4fd5a38892a8148d5aa80377db9f35ed4c4b151c'
  - 4HduymBCHhwyLgtMyXRpDX3JHQR3oyqTSytsXqCamzCc4ed9fJeBSpDUDSLwfZ59mZaw9ggdMNURPNBi4P6BRU47
x-scvd-ucp:
  profile: https://scvd.store/.well-known/ucp
  checkout: advertised
x-scvd-native-checkout:
  mcp:
    protocol: mpp
    payment_method: evm
    intent: charge
    transport: mcp
    method: tools/call
    path: /mcp
    challenge_key: org.paymentauth/payment-required
    challenge_location: error.data, or result._meta with ?payment=tool-result
    credential_meta_key: org.paymentauth/credential
    receipt_meta_key: org.paymentauth/receipt
    idempotency_meta_key: x402/idempotency-key
    terms: 'each item''s payment_capabilities row with transport http: same network, asset and amount_atomic'
  webmcp:
    protocol: mpp
    payment_method: evm
    intent: charge
    transport: webmcp
    script: /webmcp.js
    quote_tool: quote_store_purchase
    challenge_field: payment_challenge
    complete_tool: complete_store_purchase
    credential_argument: signed_credential
    receipt_field: payment_receipt
    terms: 'each item''s payment_capabilities row with transport http: same network, asset and amount_atomic'
x-rate-limiting:
  application_level_limit: true
  limited_paths:
  - /api/preflight/v1
  - /api/preflight/v2
  - /api/before-you-pay/v1
  - /api/look/v1
  - /api/preflight/batch
  headers_returned:
  - RateLimit-Limit
  - RateLimit-Remaining
  - RateLimit-Reset
  - RateLimit-Policy
  - RateLimit
  note: 'The free preflight is limited — 30 probes per isolate per minute, 60 global — because it spends outbound requests to a host the caller chooses. Every answer the limiter METERED carries the IETF RateLimit fields — the 200 and the 429 — so you can pace against the live number instead of discovering the ceiling by being refused: RateLimit-Limit / -Remaining / -Reset report whichever of the two buckets is closer to binding, and RateLimit / RateLimit-Policy name both. Past either ceiling it returns 429 with Retry-After. A validation refusal (400, e.g. a missing or unprobeable URL) returns BEFORE either bucket is touched and carries no RateLimit fields, because a malformed request never spent a probe; this contract used to declare them on those responses too, which described a header that had never been sent. No other operation enforces an application-level ceiling, and so returns no RateLimit headers: declaring a ceiling nothing enforces would be worse than declaring none. A 429 can also arrive from the edge under abuse conditions. A refused request is never charged for. The two figures above are read from the limiter''s own constants, not restated here — this string asserted that NO limit existed for a day after one shipped.'
  policy_url: https://scvd.store/developers
x-versioning:
  scheme: url-path
  note: 'Breaking changes arrive as a new version in the path (/api/preflight/v1 → /v2). A published version''s SHAPE never changes under a client: fields are added, never removed or retyped.'
  deprecation: A version being retired serves the RFC 8594 Deprecation and Sunset headers on every response for at least 90 days before it stops answering, and the date is published at /developers before the headers appear.
  sunset_headers:
  - Deprecation
  - Sunset
  - Link; rel="successor-version"
  policy_url: https://scvd.store/deprecation
  currently_deprecated: []
  versions:
  - path: /api/preflight/v1
    status: supported
    since: '2026-08-03'
    sunset: null
    successor: /api/preflight/v2
  - path: /api/preflight/v2
    status: current
    since: '2026-08-23'
    sunset: null
    successor: null
  - path: /api/look/v1
    status: current
    since: '2026-09-02'
    sunset: null
    successor: null
  - path: /api/conformance/v1
    status: current
    since: '2026-08-03'
    sunset: null
    successor: null