> ## Documentation Index > Fetch the complete documentation index at: https://docs.quackpay.com.br/llms.txt > Use this file to discover all available pages before exploring further. # Turn context into a typed decision. > Send input, an instruction and either an inline choice list or a previously created catalogue. This free endpoint does not require an API key; a bearer key is an optional alternative in the current contract. The response contains the selected choice, confidence and probabilities keyed by the original choice strings. ## OpenAPI ````yaml /openapi.json post /v1/decisions openapi: 3.1.0 info: title: Aspas Labs API version: 1.0.0 description: >- A small, direct API for structured decisions with Goldenfish. The free decision and model discovery surface does not require an API key. Send context, an instruction, and either inline choices or a reusable catalog. The response returns one selected choice, confidence, probabilities, the model used, and input-token usage. The model is selected in the request body so the public contract stays stable as the model catalog grows. Catalog management is a separate protected surface and requires a bearer key. contact: name: Aspas Labs url: https://aspaslabs.com/ email: hello@aspaslabs.com license: name: Aspas Labs API terms url: https://aspaslabs.com/ servers: - url: https://api.aspaslabs.com description: Production security: [] tags: - name: Decisions description: >- Run a constrained inference request and receive a typed choice with confidence and per-choice probabilities. - name: Models description: >- Discover model identifiers, availability status, and public input and choice limits. - name: Catalogs description: Create and manage named choice catalogs for repeatable decision workflows. - name: Health description: >- Check process liveness and whether an inference worker is ready to serve requests. paths: /v1/decisions: post: tags: - Decisions summary: Turn context into a typed decision. description: >- Send input, an instruction and either an inline choice list or a previously created catalogue. This free endpoint does not require an API key; a bearer key is an optional alternative in the current contract. The response contains the selected choice, confidence and probabilities keyed by the original choice strings. operationId: createDecision requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/DecisionRequest' examples: inlineChoices: summary: Classify a support request value: model: goldenfish-g1 input: forgot my password instruction: Which category applies? choices: - reset_password - billing - cancel_account - technical_issue catalog: summary: Use a reusable catalogue value: model: goldenfish-g1 input: billing problem instruction: What is the issue? catalog: support-actions responses: '200': description: Decision completed. headers: X-Request-ID: $ref: '#/components/headers/RequestId' content: application/json: schema: $ref: '#/components/schemas/Decision' '400': $ref: '#/components/responses/InvalidRequest' '401': $ref: '#/components/responses/AuthenticationRequired' '429': $ref: '#/components/responses/RateLimited' '503': $ref: '#/components/responses/Unavailable' security: - {} - apiKey: [] components: schemas: DecisionRequest: type: object description: >- The context and constraints for one structured decision. Exactly one of `choices` or `catalog` must be provided. additionalProperties: false required: - model - input - instruction properties: model: type: string minLength: 1 description: >- The Goldenfish model identifier. Select the model in the body instead of changing the route. examples: - goldenfish-g1 - goldenfish-g1@REVISION input: type: string minLength: 1 description: The context to classify. instruction: type: string minLength: 1 description: The question or decision instruction. choices: type: array description: >- Inline valid outcomes. Send 2–300 unique, non-empty strings when you are not using a catalog. minItems: 2 maxItems: 300 uniqueItems: true items: type: string minLength: 1 catalog: type: string minLength: 1 description: A reusable catalogue ID. oneOf: - required: - choices not: required: - catalog - required: - catalog not: required: - choices Decision: type: object description: The result of one structured decision request. required: - id - object - model - output - usage properties: id: type: string pattern: ^dec_ description: The unique decision identifier. Values begin with `dec_`. example: dec_01K7Y... object: const: decision description: Always `decision` for this response object. model: type: string description: The model identifier used to produce the result. example: goldenfish-g1 output: $ref: '#/components/schemas/DecisionOutput' usage: $ref: '#/components/schemas/Usage' DecisionOutput: type: object description: >- The selected choice and the model's probability distribution over the valid outcomes. required: - choice - confidence - probabilities properties: choice: type: string description: The selected value from the submitted choices or referenced catalog. example: reset_password confidence: type: number format: float minimum: 0 maximum: 1 description: The probability assigned to the selected choice, from 0 to 1. example: 0.9489 probabilities: type: object description: A map from every valid choice string to a probability from 0 to 1. additionalProperties: type: number format: float minimum: 0 maximum: 1 example: reset_password: 0.9489 billing: 0.0001 Usage: type: object description: Usage counters for the completed decision request. required: - input_tokens properties: input_tokens: type: integer format: int32 minimum: 0 description: Number of input tokens processed for the request. example: 13 ErrorResponse: type: object description: The common error envelope returned for failed requests. required: - error properties: error: type: object required: - message - type - code properties: message: type: string description: >- A human-readable explanation intended for logs and operator feedback. example: Rate limit exceeded. type: type: string description: The broad error category for program logic. enum: - invalid_request_error - authentication_error - rate_limit_error - model_error - server_error code: type: string description: A machine-readable reason for the failure. example: rate_limit_exceeded param: type: - string - 'null' description: The request parameter associated with the error, or `null`. example: null headers: RequestId: description: >- Correlation ID for this request. Keep it when contacting Aspas Labs about a failed or unexpected response. schema: type: string pattern: ^req_[A-Za-z0-9_-]+$ responses: InvalidRequest: description: >- The request is invalid. Check required fields, JSON types, and the rule that exactly one of `choices` or `catalog` must be provided. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' AuthenticationRequired: description: >- Authentication is required or the supplied bearer key is invalid. Add a server-side `Authorization: Bearer gf_live_...` header when the endpoint requires one. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' RateLimited: description: >- The request exceeded a rate, concurrency, or quota limit. Honor `Retry-After` when present and retry with bounded exponential backoff and jitter. headers: Retry-After: description: The minimum number of seconds to wait before retrying. schema: type: integer format: int32 minimum: 0 content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' Unavailable: description: >- The service or model is temporarily unavailable. Retry safely with bounded backoff when the operation can be repeated. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' securitySchemes: apiKey: type: http scheme: bearer bearerFormat: gf_live_... description: >- Use a bearer key beginning with `gf_live_...` for protected catalog operations. Keep keys server-side. Aspas Labs does not currently expose a self-service dashboard or key-management page. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.