Platform Backbone

Platform API, CLI & SDKs

The Platform API, CLI & SDKs are the shared control surface for rFabric. They let people, services, robots, and third-party systems act on the same resources with the same lineage, policy, and workflow history.

What This Surface Owns

Shared resource access

The platform exposes lifecycle-native resources instead of a disconnected mix of UI-only features and integration-only APIs.

Dataset, model, artifact, deployment, fleet, and incident resources
Lifecycle transitions with explicit semantics
Workflow-aware reads and writes against canonical IDs

Programmatic adoption paths

Different teams need different interfaces, but they all need to land on the same control plane.

REST and structured query surfaces
Streaming and robot-side transport where needed
CLI and SDKs for developer and operator workflows

Why It Matters

Incremental adoption without fragmentation

Teams can start with upload, release, or fleet automation without creating a separate product path from the one used by the UI.

Robot-side and human-side consistency

Robot agents, dashboards, CLI commands, workflow runners, and third-party systems all use the same resource model and permission rules.

Production-grade behavior

Versioning, validation, idempotency, auditability, and explicit state transitions matter because robotics systems do not all upgrade in lockstep.

Coexistence with the existing stack

The shared interface lets rFabric sit alongside ROS, LeRobot, visualization tooling, storage systems, and internal training environments while still becoming the system of record.

Example Lifecycle Interaction

curl https://api.rfabric.io/v1/datasets/folding_v12/finalize \
+  -H "Authorization: Bearer rf_xxx" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "curationRulesetId": "folding-production-quality",
+    "requiredAnnotations": ["phase_labels", "success_failure"],
+    "splits": {"train": 0.88, "val": 0.08, "test": 0.04},
+    "exportFormats": ["lerobot_v3", "parquet_manifest"]
+  }'

The important point is not the exact syntax. It is that this action operates on a canonical dataset object with explicit policy, lineage, and downstream meaning shared across UI, CLI, SDKs, and workflow runtime.

Why Teams Care

Adoption fit

Teams can bring rFabric into their stack without rewriting everything behind one monolithic web application.

Consistency

UI, CLI, SDK, and automation all speak the same entity language and lifecycle state model.

Reliability

Stable contracts and explicit validation reduce breakage in robot-side and operations-critical integrations.

Scale

The platform can extend into CI, training repos, edge agents, customer systems, and internal tooling while preserving one control plane.