openapi: 3.0.3 info: title: Repzo API - AI Object Detection Task version: 1.0.0 description: | **Object Detection Tasks** are the atomic annotatable unit of the AI object-detection pipeline. Each task is one image plus its `annotation_groups[]` — versioned sets of 2D bounding-box `annotations[]` (each with a `box`, `confidence`, `label_id`, and `label_state`). Tasks feed datasets, model-version training, and the dashboard review canvas. **Sessions (AR shelf scans).** A task may optionally belong to an `ai-object-detection-session` (set via the `session` field). When it does, it also carries per-frame AR context in `frame_meta` (pose, intrinsics, tracking, depth source), raw `depth_media` / `confidence_media` blobs, and `depth_shape`. When such a frame is inferred (via `ai-object-detection-inference` with a `model_version`, directly or driven by `ai-object-detection-session-analysis`), each placeable detection is back-projected into world coordinates: individual `annotations[]` gain `world_position`, `world_size`, `depth_at_center`, `depth_confidence`, and `placement_confidence` (geometry only — the kept/ignored/merged conclusion is made by the analysis stage, not stored here). Human-made / edited boxes (`manual` / `auto_edited` groups) get the same placement at SAVE time: a box whose geometry no longer matches its `placed_box` snapshot is re-placed on the next `PUT`. After a successful analysis run, each contributing annotation's `cluster_id` points at the concluded object (`objects[]._id`) on the `ai-object-detection-session-analysis` document. All of these fields are additive and absent on tasks created from the dashboard upload flow. **Who calls it.** Admins and reps from the dashboard (upload, review, bulk-approve). Session frames are materialized as tasks by the `activity-ai-object-detection-session-frame` intake service. **Multi-tenancy & lifecycle.** Records are scoped by `company_namespace[]` (server-injected on create / update). Soft-delete via `disabled: true` (DELETE flips the flag). `annotated` is derived on write: `true` when any confirmed group has annotations. `creator` is stamped on create, `editor` on update. A `PUT` that really changes a session frame's human annotations (moved / new boxes, adds / deletes, re-labels, a fresh group) schedules a re-analysis of the parent session; flag flips do not. **Key relationships.** References `media-storage` (`file_media`, `depth_media`, `confidence_media`, XAI images), `ai-object-detection-label` (`annotations.label_id`, `original_label`), `ai-object-detection-model-version` (`annotation_groups.model_version`, `annotated_by_model_version_code`), `ai-object-detection-dataset` (`task_dataset.dataset`), and `ai-object-detection-session` (`session`). **Bulk patch.** `PATCH /ai-object-detection-task` (no id) accepts a `patch-action`-style `writeQuery` and applies it to all tasks matching the query filters (the same keys as find, e.g. `?session=:id` to approve all auto-annotations for a session). Returns `{ nFound, nModified }`. **Population.** `?populatedKeys[]=` supports `file_media`, `task_dataset.dataset`, `annotation_groups.model_version`, `annotation_groups.annotations.label_id`, `annotation_groups.annotations.original_label`, `annotation_groups.annotated_by_model_version_code`, `session`, `depth_media`, `confidence_media`, `annotation_groups.xai.heatmap_media`, `annotation_groups.xai.gradcam_media`, `annotation_groups.xai.gradcam_per_detection.media`, and `annotation_groups.xai.feature_maps.media`. TOP-LEVEL refs (`file_media`, `session`, `depth_media`, `confidence_media`) are emitted under `_populated` while the ref itself stays an id; every NESTED path (under `task_dataset.` / `annotation_groups.`) populates INLINE, replacing the id with the document. servers: - url: https://sv.api.repzo.me security: - ApiKeyAuth: [] - JwtAuth: [] paths: /ai-object-detection-task: get: summary: Find object detection tasks operationId: findAiObjectDetectionTasks parameters: - in: query name: _id description: | Filter by task `_id`. Pass once for a single match, or as `?_id[]=...&_id[]=...` for multiple. schema: oneOf: - type: string - type: array items: { type: string } - in: query name: session description: Filter tasks (frames) belonging to a session `_id`. schema: oneOf: - type: string - type: array items: { type: string } - in: query name: frame_meta.frame_id description: Filter by the session-scoped frame counter. schema: oneOf: - type: number - type: array items: { type: number } - in: query name: task_dataset.dataset description: Filter tasks belonging to a dataset `_id`. schema: oneOf: - type: string - type: array items: { type: string } - in: query name: annotated description: Filter by whether the task has any confirmed annotations. schema: { type: boolean } - in: query name: annotated_label description: | Only tasks containing at least one annotation (in any group) whose `label_id` is one of the given label `_id`s. Used to surface candidate crops for a label's reference photo. Non-ObjectId values are ignored. schema: oneOf: - type: string - type: array items: { type: string } - in: query name: annotation_status description: | Filter by the task's mutually-exclusive annotation workflow status. The filter is applied before counting and pagination. `pending` means the task is not annotated (or has no annotation groups); `confirmed` means its leading annotation group is confirmed; `manual` and `auto_edited` select annotated tasks whose leading group is unconfirmed with that state; `auto` selects the remaining annotated tasks with an unconfirmed leading group. Any other value is rejected with `400`. schema: type: string enum: [pending, confirmed, manual, auto_edited, auto] - in: query name: disabled description: "Include soft-deleted tasks. Omitted / `false` returns active tasks only; `true` returns disabled tasks only." schema: { type: boolean, default: false } - in: query name: from_createdAt description: Return tasks created at/after this Unix timestamp (ms) or date string. schema: { type: number } - in: query name: to_createdAt description: Return tasks created at/before this Unix timestamp (ms) or date string. schema: { type: number } - in: query name: from_updatedAt description: Cursor — tasks with `updatedAt` at/after this Unix timestamp (ms). schema: { type: number } - in: query name: to_updatedAt description: Cursor — tasks with `updatedAt` at/before this Unix timestamp (ms). schema: { type: number } - in: query name: populatedKeys description: | Embed referenced documents. Top-level refs land under `_populated`; nested paths populate inline (see the intro). schema: type: array items: type: string enum: - file_media - task_dataset.dataset - annotation_groups.model_version - annotation_groups.annotations.label_id - annotation_groups.annotations.original_label - annotation_groups.annotated_by_model_version_code - session - depth_media - confidence_media - annotation_groups.xai.heatmap_media - annotation_groups.xai.gradcam_media - annotation_groups.xai.gradcam_per_detection.media - annotation_groups.xai.feature_maps.media - in: query name: per_page description: Page size. Defaults to the server's configured pagination limit. schema: { type: integer, minimum: 1, maximum: 50000 } example: 50 - in: query name: page description: 1-indexed page number. schema: { type: integer, minimum: 1 } example: 1 - in: query name: sort description: "Field to sort by. Defaults to `_id`." schema: { type: string, default: _id } - in: query name: sortPageOrder description: "Sort direction. Defaults to descending (`dsc`)." schema: { type: string, enum: [asc, dsc], default: dsc } responses: "200": description: Paginated list of object detection tasks. content: application/json: schema: $ref: "#/components/schemas/TaskFindResult" "400": description: "Invalid `annotation_status` value." post: summary: Create an object detection task description: | Creates a task. `company_namespace` and `creator` are server-injected. Each `annotation_group` MUST carry a valid `annotation_state` (else `400`): `manual` / `auto_edited` mark the group `confirmed` (+ `confirmed_by` from the JWT), `auto` records `annotated_by_model_version_code` = `model_version`. A missing `time` defaults to now. `annotated` is derived. operationId: createAiObjectDetectionTask requestBody: required: true content: application/json: schema: $ref: "#/components/schemas/TaskCreateBody" responses: "201": description: The newly-created task document. content: application/json: schema: $ref: "#/components/schemas/TaskSchema" "400": description: "An annotation group has a missing / invalid `annotation_state`, or model validation failed (e.g. missing `file_media`)." patch: summary: Bulk-update tasks description: | Bulk update via the `patch-action` `writeQuery` shape, applied to all tasks matching the query filters (the same accepted keys as find — no pagination / population / `annotation_status`). Used for session-scoped bulk actions (e.g. `?session=:id` with a writeQuery that sets annotation groups confirmed). Returns `{ nFound, nModified }`. operationId: patchAiObjectDetectionTask parameters: - in: query name: _id description: Restrict the bulk update to these task `_id`s. schema: oneOf: - type: string - type: array items: { type: string } - in: query name: session description: Restrict the bulk update to tasks of this session `_id`. schema: oneOf: - type: string - type: array items: { type: string } - in: query name: frame_meta.frame_id description: Restrict to the given session-scoped frame counter(s). schema: oneOf: - type: number - type: array items: { type: number } - in: query name: task_dataset.dataset description: Restrict to tasks belonging to a dataset `_id`. schema: oneOf: - type: string - type: array items: { type: string } - in: query name: annotated schema: { type: boolean } - in: query name: disabled schema: { type: boolean, default: false } - in: query name: from_createdAt schema: { type: number } - in: query name: to_createdAt schema: { type: number } - in: query name: from_updatedAt schema: { type: number } - in: query name: to_updatedAt schema: { type: number } requestBody: required: true content: application/json: schema: $ref: "#/components/schemas/PatchActionBody" responses: "200": description: Bulk-update result. content: application/json: schema: $ref: "#/components/schemas/PatchActionResult" "400": description: "Unknown `writeQuery[].command`." /ai-object-detection-task/{id}: get: summary: Get a task by id operationId: getAiObjectDetectionTask parameters: - in: path name: id required: true schema: { type: string } - in: query name: populatedKeys description: Same keys as on find. schema: type: array items: type: string enum: - file_media - task_dataset.dataset - annotation_groups.model_version - annotation_groups.annotations.label_id - annotation_groups.annotations.original_label - annotation_groups.annotated_by_model_version_code - session - depth_media - confidence_media - annotation_groups.xai.heatmap_media - annotation_groups.xai.gradcam_media - annotation_groups.xai.gradcam_per_detection.media - annotation_groups.xai.feature_maps.media responses: "200": description: The task document for the given `_id`. content: application/json: schema: $ref: "#/components/schemas/TaskSchema" "400": description: No task with that `_id` in the caller's namespace (note — 400, not 404). put: summary: Update a task description: | Standard put — typically to edit annotations during review. The body's `annotation_groups` REPLACES the stored array wholesale, so clients must resend every group they want to keep; revising an existing group means resending it edited in place (not appending a copy). `editor` is server-injected; `annotated` is re-derived; `edit_time` is restamped on every group; an `auto_edited` group is marked `confirmed`; each confirmed group's `confirmed_by` is stamped from the JWT only when it doesn't already carry a valid user — the original confirmer is preserved across re-saves. On session frames, human boxes (`manual` / `auto_edited` groups) whose geometry changed since their `placed_box` are re-placed in world coordinates (best-effort — a failure records `ignore_reason` and never blocks the save), and a real content change schedules a session re-analysis. operationId: updateAiObjectDetectionTask parameters: - in: path name: id required: true schema: { type: string } requestBody: required: true content: application/json: schema: $ref: "#/components/schemas/TaskUpdateBody" responses: "200": description: The task document after the update is applied. content: application/json: schema: $ref: "#/components/schemas/TaskSchema" "404": description: No task with that `_id` in the caller's namespace. delete: summary: Soft-delete a task description: "Sets `disabled: true` and returns the task." operationId: removeAiObjectDetectionTask parameters: - in: path name: id required: true schema: { type: string } responses: "200": description: "The task document after soft-deletion (`disabled: true`)." content: application/json: schema: $ref: "#/components/schemas/TaskSchema" "400": description: No task with that `_id` in the caller's namespace. components: securitySchemes: ApiKeyAuth: type: apiKey in: header name: api-key description: | Server-issued API key. Also accepted via the `x-api-key` header or the `?apiKey=` query parameter as fallbacks. JwtAuth: type: apiKey in: header name: Authorization description: | Raw JWT in the `Authorization` header — **no `Bearer ` prefix**. Obtained from `POST /authenticate` (admin / rep / client login). schemas: UserRef: type: object properties: _id: { type: string } type: { type: string, enum: [admin, rep] } name: { type: string } rep: { type: string } admin: { type: string } Box: type: object description: "Normalized box. Inference and the dashboard store YOLO center-size here: `x1` = cx, `y1` = cy, `x2` = w, `y2` = h." required: [x1, x2, y1, y2] properties: x1: { type: number } x2: { type: number } y1: { type: number } y2: { type: number } Annotation: type: object required: [box, label_id, label_state] properties: _id: { type: string } box: $ref: "#/components/schemas/Box" confidence: { type: number, default: 0.25 } label_id: type: string description: "`ai-object-detection-label` id." label_state: { type: string, enum: [auto, manual] } placed_box: allOf: - $ref: "#/components/schemas/Box" description: | The box geometry world placement was last computed FOR. Set by the placement pass (inference for auto groups; the task-save pass for manual/auto_edited groups). When `box` differs from `placed_box` (a human moved/redrew it), the next save recomputes the world fields — so every annotation carries AR data from birth. world_position: type: object description: | Back-projected centroid in world coordinates, **metres**. Written by the PLACEMENT pass: at inference time for model detections, and at task-save time for human-made/edited boxes (manual / auto_edited groups on session frames with usable pose/intrinsics/depth). properties: x: { type: number } y: { type: number } z: { type: number } world_size: type: object description: "Physical front-face size of the detection, **centimetres** (`w` = world-horizontal, `h` = vertical)." properties: w: { type: number } h: { type: number } depth_at_center: type: number description: | Distance from the camera, **metres** (session frames only) — the front-biased percentile (default p30) of the depth pixels sampled over the whole detection bbox. depth_confidence: type: number description: Normalized 0..1 depth confidence (session frames only). placement_confidence: type: number description: "detection_conf x depth_conf x tracking (placed session frames only)." cluster_id: type: string description: | Link to the analysis object this detection was clustered into. Note: the kept/ignored/merged conclusion lives on the `ai-object-detection-session-analysis` doc, **not** on the annotation. ignore_reason: type: string enum: - no_pose - no_intrinsics - no_depth - empty_depth_region - insufficient_depth_pixels - behind_shelf description: | Why this detection could NOT be placed in world coordinates (set by the placement pass; unset when placed). Frame-level causes: `no_pose` / `no_intrinsics` / `no_depth`. Per-box causes: `empty_depth_region` (box maps to no depth pixels), `insufficient_depth_pixels` (fewer than 3 valid pixels), `behind_shelf` (only background depth beyond the shelf gate). original_label: type: string description: | Dims-reclassifier provenance — when the reclassifier moves this annotation to a sibling label, the FIRST original label is kept here (written once; cleared if a later run reverts to it). reclassification_reason: type: string enum: [dims_match_sibling, group_consensus, manual] size_rejected: type: boolean description: | Size-gate provenance — set when the measured physical size exceeded the label's expected dims beyond the allowance and the detection was dropped from the analysis walk. The annotation is KEPT (never deleted); cleared when a later run passes it. size_reject_detail: type: object properties: exceeded: { type: string, enum: [width, height, area] } measured_w_cm: { type: number } measured_h_cm: { type: number } expected_w_cm: { type: number } expected_h_cm: { type: number } ratio: type: number description: measured / (expected × (1 + allowance)) for the tripped check. AnnotationGroup: type: object properties: _id: { type: string } model_version: { type: string } engine: type: string enum: [trained, zero_shot] description: | Which detector produced this group: a trained model version (default) or a zero-shot VLM (no trained model yet). zero_shot_model: type: string description: "The zero-shot VLM used, e.g. `qwen/qwen3-vl-8b-instruct`." usable: type: boolean default: true description: Opt-out training flag. time: type: number description: Epoch ms; defaults to now on create. edit_time: type: number description: Epoch ms; server-stamped on every update. annotation_state: { type: string, enum: [auto, manual, auto_edited] } confirmed: { type: boolean, default: false } confirmed_by: $ref: "#/components/schemas/UserRef" annotated_by_model_version_code: { type: string } annotations: type: array items: $ref: "#/components/schemas/Annotation" inference_run: type: string description: Link to the session's `inference_runs[]._id` (session inference only). session_inference: type: object description: Per-group session inference metadata (session inference only). properties: plane_normal: type: array items: { type: number } is_winner_in_cluster: { type: boolean } cross_frame_overlap_pct: { type: number } xai: $ref: "#/components/schemas/GroupXai" AnnotationGroupCreate: description: "An annotation group as sent on create — `annotation_state` is REQUIRED (a group without a valid one is rejected with 400). Server-stamped fields (`confirmed_by`, `edit_time`, `annotated_by_model_version_code`) are ignored." allOf: - $ref: "#/components/schemas/AnnotationGroup" - type: object required: [annotation_state] GroupXai: type: object description: | Explainable-AI results persisted by the inference service when the group was produced with `explain: true`. Heatmap / feature-map images live in media storage — each ref is paired with a `*url` publicUrl snapshot so consumers can render without a populate round-trip (`populatedKeys[]=annotation_groups.xai.heatmap_media` and `annotation_groups.xai.feature_maps.media` populate the refs inline). `class_scores` / `embeddings.points` reference the lambda predictions by `index` (an unmapped prediction never becomes an annotation, so they cannot key to `annotations[]`). properties: parts: type: array items: { type: string } description: Which XAI parts were requested when the group was made. heatmap_media: type: string description: Media ref of the EigenCAM RGBA PNG (alpha = activation). heatmap_url: type: string description: publicUrl snapshot of `heatmap_media`. heatmap_method: { type: string, enum: [eigencam] } heatmap_width: { type: integer } heatmap_height: { type: integer } gradcam_media: type: string description: "Media ref of the TRUE gradient Grad-CAM RGBA PNG (alpha = activation)." gradcam_url: type: string description: publicUrl snapshot of `gradcam_media`. gradcam_width: { type: integer } gradcam_height: { type: integer } gradcam_per_detection: type: array description: | Per-detection Grad-CAM maps ("why THIS object") — one image per top-K detection, same overlay contract as `gradcam_media`. `index` references the lambda's predictions like `class_scores`. Populate the refs with `annotation_groups.xai.gradcam_per_detection.media`. items: type: object properties: index: { type: integer } class: { type: integer } name: { type: string } confidence: { type: number } box: type: object description: Detection box in original-image pixels (xyxy). properties: x1: { type: number } y1: { type: number } x2: { type: number } y2: { type: number } match_iou: { type: number } media: type: string description: Media ref of this detection's RGBA PNG map. url: type: string description: publicUrl snapshot of `media`. width: { type: integer } height: { type: integer } feature_maps: type: array items: type: object properties: stage: type: string description: "Network stage, e.g. `stage12_C2f`." media: type: string description: Media ref of the stage grid JPEG. url: type: string description: publicUrl snapshot of `media`. class_scores: type: array description: Per-detection top-k class scores (see the inference spec). items: type: object properties: index: { type: integer } class: { type: integer } name: { type: string } confidence: { type: number } box: type: object properties: x1: { type: number } y1: { type: number } x2: { type: number } y2: { type: number } match_iou: { type: number } candidates: type: array items: type: object properties: class: { type: integer } name: { type: string } score: { type: number } embeddings: type: object description: 2D projection of per-detection embeddings. properties: method: type: string enum: [umap, tsne, pca, none] description: | The projection that actually ran. On `auto` — UMAP for 10+ detections, t-SNE for 5-9, PCA for 2-4, `none` for a single point; explicit requests degrade down the umap -> tsne -> pca chain when infeasible. requested_method: type: string enum: [auto, umap, tsne, pca] description: "What the caller asked for (differs from `method` on fallback)." layer: { type: string } embedding_dim: { type: integer } points: type: array items: type: object properties: index: { type: integer } class: { type: integer } name: { type: string } confidence: { type: number } x: { type: number } y: { type: number } confusion_matrix: type: object description: Training-time confusion-matrix image URLs of the model version. properties: url: { type: string } normalized_url: { type: string } source: { type: string, enum: [training_artifacts] } notes: type: array items: { type: string } description: Reasons for any XAI part that could not be produced or persisted. computed_at: type: number description: Epoch ms when the XAI results were persisted (withheld when no image could be stored). TaskDataset: type: object required: [dataset, subset] properties: _id: { type: string } dataset: { type: string } subset: { type: string, enum: [train, val, test, auto, ignore] } FrameMeta: type: object description: Per-frame AR context, present only on session frames. properties: frame_id: { type: number } ts: { type: number } pose: type: array items: { type: number } description: 16-float 4x4 pose matrix, column-major. euler_ypr: type: array items: { type: number } intrinsics: type: object properties: fx: { type: number } fy: { type: number } cx: { type: number } cy: { type: number } distortion: type: object properties: k1: { type: number } k2: { type: number } k3: { type: number } p1: { type: number } p2: { type: number } tracking: type: object properties: state: { type: string, enum: [NORMAL, LIMITED, LOST] } score: { type: number } drift_m: { type: number } velocity_mps: { type: number } distance_to_shelf_m: type: number description: Shelf-distance gate used when back-projecting this frame's detections. image_rotation_deg: type: number description: CW degrees the sensor image was rotated to produce the stored image (0/90/180/270). image_stats: type: object properties: iso: { type: number } shutter: { type: number } lux: { type: number } sharpness: type: number description: Variance-of-Laplacian focus measure (higher = sharper). depth_summary: type: object description: Depth-map presence + confidence fractions (election input). properties: width: { type: number } height: { type: number } conf_high: { type: number } conf_medium: { type: number } conf_low: { type: number } depth_source: { type: string, enum: [lidar, estimated, none] } min_distance_depth: type: number description: Nearest depth sample in the frame, meters. max_distance_depth: type: number description: Farthest depth sample in the frame, meters. depth_variation: type: number description: Depth span (max - min), meters. yaw_degree: type: number description: Camera yaw at capture, degrees. pitch_degree: type: number description: Camera pitch at capture, degrees. roll_degree: type: number description: Camera roll at capture, degrees. frame_sharpness: type: number description: Variance-of-Laplacian focus measure (higher = sharper; unmeasured frames omit it). capture_tier: type: string enum: [good, warn, error] description: Device validator verdict for the frame at capture time. detection_settings_polled_at: type: number description: Epoch ms of the app's last successful detection-settings poll before this frame was submitted. app_version: type: string description: App build that captured the frame. TaskSchema: type: object description: Object detection task document. properties: _id: { type: string } file_media: { type: string } shape: type: array items: { type: number } annotation_groups: type: array items: $ref: "#/components/schemas/AnnotationGroup" task_dataset: type: array items: $ref: "#/components/schemas/TaskDataset" creator: $ref: "#/components/schemas/UserRef" editor: $ref: "#/components/schemas/UserRef" annotated: { type: boolean } session: type: string description: Parent session `_id` (session frames only). frame_meta: $ref: "#/components/schemas/FrameMeta" depth_media: type: string description: Media id of the raw float32 depth blob (session frames only). confidence_media: type: string description: Media id of the raw uint8 confidence blob (session frames only). depth_shape: type: array items: { type: number } description: "[width, height] of the depth blob." disabled: { type: boolean } company_namespace: type: array items: { type: string } description: Tenant key. Server-injected — never accept from clients. createdAt: { type: string, format: date-time } updatedAt: { type: string, format: date-time } file_media_populated: type: object description: "The `media-storage` document, when `populatedKeys[]=file_media` was requested." session_populated: type: object description: "The `ai-object-detection-session` document, when `populatedKeys[]=session` was requested." depth_media_populated: type: object description: "The `media-storage` document, when `populatedKeys[]=depth_media` was requested." confidence_media_populated: type: object description: "The `media-storage` document, when `populatedKeys[]=confidence_media` was requested." TaskCreateBody: type: object description: | Body for creating a task. `creator` is server-injected — do not send it. The tenant key (`company_namespace`) is optional for SDK callers and is otherwise injected from the caller's session. required: [file_media] properties: file_media: { type: string } shape: type: array items: { type: number } annotation_groups: type: array items: $ref: "#/components/schemas/AnnotationGroupCreate" task_dataset: type: array items: $ref: "#/components/schemas/TaskDataset" session: { type: string } frame_meta: $ref: "#/components/schemas/FrameMeta" depth_media: { type: string } confidence_media: { type: string } depth_shape: type: array items: { type: number } company_namespace: type: array items: { type: string } description: Optional tenant namespace override for SDK callers. TaskUpdateBody: type: object description: | Body for updating a task. `editor` is server-injected and `annotated` re-derived. `annotation_groups` replaces the stored array wholesale. Set `disabled: true` to soft-delete. properties: file_media: { type: string } shape: type: array items: { type: number } annotation_groups: type: array items: $ref: "#/components/schemas/AnnotationGroup" task_dataset: type: array items: $ref: "#/components/schemas/TaskDataset" session: { type: string } frame_meta: $ref: "#/components/schemas/FrameMeta" depth_media: { type: string } confidence_media: { type: string } depth_shape: type: array items: { type: number } disabled: { type: boolean } PatchActionWrite: type: object required: [key, command, value] properties: key: type: string description: "Dotted document path, e.g. `annotation_groups.0.confirmed`." command: type: string enum: [set, addToSet, pull] description: "`set` → `$set`; `addToSet` → `$addToSet: { $each: value }`; `pull` → `$pull: { $in: value }`." value: {} PatchActionBody: type: object description: Bulk-update body. `writeQuery[]` describes the writes applied to filtered rows. required: [writeQuery] properties: writeQuery: type: array items: $ref: "#/components/schemas/PatchActionWrite" PatchActionResult: type: object properties: nFound: type: number description: Matched documents. nModified: type: number description: Modified documents. TaskFindResult: type: object description: Standard paginated result envelope. properties: data: type: array items: $ref: "#/components/schemas/TaskSchema" total_result: { type: number } current_count: { type: number } total_pages: { type: number } current_page: { type: number } per_page: { type: number } first_page_url: { type: string } last_page_url: { type: string } next_page_url: { type: string, nullable: true } prev_page_url: { type: string, nullable: true } path: { type: string }