API Evangelist conversational phrasing for Edge Impulse Classify API

# Generated by API Evangelist (build-phrasing.py). Our phrasing, not observed demand.
overlay: 1.0.0
info:
  title: API Evangelist conversational phrasing for Edge Impulse Classify API
  version: 1.0.0
extends: openapi/edge-impulse-classify-api-openapi.yml
actions:
- target: $.info
  update:
    x-apievangelist-phrasing:
      method: generated
      generated: '2026-10-01'
      generator: build-phrasing.py
      label: Generated by API Evangelist
      operations: 10
- target: $.paths['/api/{projectId}/classify/{sampleId}'].get
  update:
    x-apievangelist-phrasing:
      intent: Classify a sample (deprecated v1)
      effect: read
      questions:
      - Is the old GET classify-sample endpoint still usable, or has it been deprecated?
      - Can the legacy v1 classifier return debug info for a whole file?
      instructions:
      - text: Classify sample {sampleId} with the deprecated v1 classify endpoint.
        slots:
          sampleId: path.sampleId
      - text: Use the legacy v1 classifier on sample {sampleId} in project {projectId} against impulse {impulseId}.
        slots:
          sampleId: path.sampleId
          projectId: path.projectId
          impulseId: query.impulseId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/v2/{sampleId}'].post
  update:
    x-apievangelist-phrasing:
      intent: Classify a sample against the impulse
      effect: write
      questions:
      - How do I run my trained impulse over a complete sample file with a sliding window?
      - Can I classify a sample with a specific model variant, like the quantized one?
      - Why might classifying a sample return a job ID instead of results?
      instructions:
      - text: Classify sample {sampleId} in project {projectId} with the current impulse.
        slots:
          sampleId: path.sampleId
          projectId: path.projectId
      - text: Classify sample {sampleId} using the {variant} model variant and include debug info.
        slots:
          sampleId: path.sampleId
          variant: query.variant
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/v2/{sampleId}/variants'].post
  update:
    x-apievangelist-phrasing:
      intent: Classify a sample across several model variants
      effect: write
      questions:
      - Can I compare how float32 and int8 variants classify the same sample in one call?
      - How do I classify one sample against multiple model variants at once?
      instructions:
      - text: Classify sample {sampleId} for the variants {variants}.
        slots:
          sampleId: path.sampleId
          variants: query.variants
      - text: Run sample {sampleId} in project {projectId} through every variant in {variants} side by side.
        slots:
          sampleId: path.sampleId
          projectId: path.projectId
          variants: query.variants
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/v2/{sampleId}/raw-data/{windowIndex}'].get
  update:
    x-apievangelist-phrasing:
      intent: Get cached raw features for a sample window
      effect: read
      questions:
      - After live classification finishes, can I fetch the raw features of one window?
      - Which window of a classified sample can I pull from the feature cache?
      instructions:
      - text: Get cached raw features for window {windowIndex} of sample {sampleId}.
        slots:
          windowIndex: path.windowIndex
          sampleId: path.sampleId
      - text: Pull window {windowIndex} of sample {sampleId} in project {projectId} from the classification cache.
        slots:
          windowIndex: path.windowIndex
          sampleId: path.sampleId
          projectId: path.projectId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/all/result'].get
  update:
    x-apievangelist-phrasing:
      intent: Get the full test-set classification result
      effect: read
      questions:
      - Where do I see how my model performed across the whole testing dataset?
      - Can I fetch only the feature explorer data from the classify-all job?
      instructions:
      - text: Get the classify-all job result for project {projectId}.
        slots:
          projectId: path.projectId
      - text: Show the complete test-set results for the {variant} variant.
        slots:
          variant: query.variant
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/all/result/page'].get
  update:
    x-apievangelist-phrasing:
      intent: Page through test-set predictions
      effect: read
      questions:
      - How do I page through the predictions from a classify-all job?
      - Can I limit how many test-set predictions come back per page?
      instructions:
      - text: Get page of {limit} test predictions starting at {offset} for project {projectId}.
        slots:
          limit: query.limit
          offset: query.offset
          projectId: path.projectId
      - text: Fetch the next page of classify-all predictions from offset {offset}.
        slots:
          offset: query.offset
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/all/metrics'].get
  update:
    x-apievangelist-phrasing:
      intent: Get test metrics for every model variant
      effect: read
      questions:
      - What accuracy did each model variant reach in the last classify-all run?
      - Can I compare test metrics across all variants of my impulse?
      instructions:
      - text: Get classify-all metrics for every model variant in project {projectId}.
        slots:
          projectId: path.projectId
      - text: Compare test metrics for all variants of impulse {impulseId}.
        slots:
          impulseId: query.impulseId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/anomaly-gmm/{blockId}/{sampleId}'].get
  update:
    x-apievangelist-phrasing:
      intent: Classify a sample by learn block (deprecated v1)
      effect: read
      questions:
      - Is the v1 anomaly GMM learn-block classify endpoint deprecated?
      - What did the older GET learn-block classifier return for a sample?
      instructions:
      - text: Classify sample {sampleId} against learn block {blockId} with the deprecated v1 endpoint.
        slots:
          sampleId: path.sampleId
          blockId: path.blockId
      - text: Use the legacy v1 anomaly classifier for block {blockId} on sample {sampleId} in project {projectId}.
        slots:
          blockId: path.blockId
          sampleId: path.sampleId
          projectId: path.projectId
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/anomaly-gmm/v2/{blockId}/{sampleId}'].post
  update:
    x-apievangelist-phrasing:
      intent: Classify a sample with one learn block
      effect: write
      questions:
      - How do I run a single learn block, like the anomaly GMM, over a whole sample?
      - Can I score a sample against just one learn block instead of the full impulse?
      instructions:
      - text: Classify sample {sampleId} with learn block {blockId} in project {projectId}.
        slots:
          sampleId: path.sampleId
          blockId: path.blockId
          projectId: path.projectId
      - text: Run anomaly learn block {blockId} on sample {sampleId} using the {variant} variant.
        slots:
          blockId: path.blockId
          sampleId: path.sampleId
          variant: query.variant
      method: generated
      generated: '2026-10-01'
- target: $.paths['/api/{projectId}/classify/image'].post
  update:
    x-apievangelist-phrasing:
      intent: Classify an uploaded image
      effect: write
      questions:
      - Can I test my trained image model by posting a photo directly?
      - How do I classify an image that isn't in my dataset?
      instructions:
      - text: Classify image {image} with the trained impulse in project {projectId}.
        slots:
          image: requestBody.image
          projectId: path.projectId
      - text: Test impulse {impulseId} on the image {image}.
        slots:
          impulseId: query.impulseId
          image: requestBody.image
      method: generated
      generated: '2026-10-01'