RTK Motion — Motion Capture, Biomechanics, Threat & Fleet Intelligence

RTK Motion Intelligence — MCP Server

Agent-to-agent paid data marketplace. Two-person (dyadic) motion capture in a shared coordinate frame, rehab biomechanics, threat intelligence, and commercial-carrier truck-sighting intelligence (RTK Spotter). 21 MCP tools, gasless USDC payments via x402 (EIP-3009).

rtkmotion.io · Developer docs · Live catalog · Pricing · Shared-frame mocap spec

What is this?

A live, hosted Model Context Protocol server providing agent-to-agent paid access to:

• 4D Motion Capture — BVH skeletal animation, multi-view video, robot-ready joint trajectories (NPZ). Two-person (dyadic) capture in a shared coordinate frame: both actors' global root translation and orientation are preserved in one capture-volume frame, so contact, separation and relative pose are recoverable directly from the files. Most released datasets canonicalize each actor to its own origin, which destroys inter-actor geometry irreversibly — see the specification and its pre-purchase verification procedure
• Rehabilitation Biomechanics — cross-exercise summaries, per-exercise reports, ROM and asymmetry metrics
• Location Threat Intelligence — confidence-scored alerts, event distribution, temporal context
• RTK Spotter — Commercial Vehicle Intelligence — de-identified commercial-carrier truck sightings (per-USDOT bundles) plus a geo-stripped truck-recognition CV/OCR training corpus. Commercial carriers only; faces + bystander plates blurred; FMCSA-public labels (USDOT, carrier, equipment)

The server runs at https://api.rtkmotion.io/mcp on Cloudflare Workers, using the streamable-HTTP MCP transport.

This repository is the public-facing manifest for catalog discovery. The server itself is hosted, not installable from this repo.

Quick start

Add to Claude Desktop / Cursor / Cline / Continue

Add to your client's MCP config (e.g. ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "rtk-motion": {
      "type": "streamable-http",
      "url": "https://api.rtkmotion.io/mcp"
    }
  }
}

Restart your client. The 21 tools become available.

See examples/ for ready-to-use config snippets and a minimal Python connect script.

Free tools (no payment, 10 total)

• browse_catalog — hierarchical catalog of all data
• search_catalog — full-text search across motion capture, rehab, threat, and Spotter datasets
• get_mocap_sample — first 5 s of BVH skeletal animation (300 frames at 60 fps)
• get_calibration — multi-camera capture-rig calibration
• get_threat_profile — monitored location profile (assets, feeds, scan interval)
• get_threat_sample — current threat assessment sample
• get_rehab_profile — patient demographics, injury history, medications
• get_rehab_sample — rehab biomechanics sample
• get_spotter_carrier_profile — commercial-carrier coverage + de-identification posture
• get_spotter_cv_profile — truck CV/OCR corpus profile (image count, label schema)

Paid tools (USDC via x402, 11 total)

Tool	Price	Output
get_bvh	$10	Full BVH skeletal animation
get_video_url	$5	Multi-view grappling video
get_robotarget	$15	Robot-ready joint trajectories (NPZ, MoveIt/ROS/Isaac compatible)
get_trial_bvh	$0.25	Trial BVH slice (5 s) — evaluate quality
get_trial_robotarget	$0.25	Trial NPZ trajectory (5 s) — evaluate quality
get_threat_summary	$0.50 – $1.00	Location threat assessment (lite / full)
get_rehab_summary	$0.50	Cross-exercise biomechanical summary
get_rehab_report	$0.25	Per-exercise biomechanical report
get_spotter_carrier	$5	Per-USDOT commercial-carrier sighting bundle (de-identified crops + trail)
get_spotter_cv_sample	$0.25	Trial slice of the truck CV/OCR corpus
get_spotter_cv	$50	Full geo-stripped truck-recognition CV/OCR training corpus

Payment in USDC via the x402 payment protocol. Which networks are available depends on the flow:

Flow	Networks	How
Gasless x402 exact / EIP-3009 (preferred)	Base	Agent signs a single-use transfer authorization and passes it as x_payment — no gas for the buyer
On-chain payment_tx (legacy, MCP tools)	Base · Ethereum · Solana	Send USDC, then supply the settled transaction hash
HTTP /paid/* routes	Base only	CDP-facilitated; the 402 challenge advertises eip155:8453

Full pricing detail at api.rtkmotion.io/pricing.

Provenance

Every paid record is signed with ECDSA (secp256k1) by 0x6C11F8a21f7ca922F483Ed21C3b6c2d9B305B10C. Verification key (JWKS) at api.rtkmotion.io/.well-known/jwks.json.

Discovery surfaces

Surface	URL
Website	rtkmotion.io · developer docs
Glama connector	glama.ai/mcp/servers/RTKMotion/motion-intelligence
Smithery catalog	smithery.ai/servers/RTKMotion/motion-intelligence
MCP Server Registry	io.rtkmotion/data (listing)
A2A Agent Card	api.rtkmotion.io/.well-known/agent.json
x402 Manifest	api.rtkmotion.io/.well-known/x402
OpenAPI 3.1	api.rtkmotion.io/openapi.json
Live data catalog (free)	api.rtkmotion.io/catalog

Use cases

• Robotics — Motion capture for sim-to-real kinematic retargeting (MoveIt/ROS/Isaac compatible NPZ trajectories). Dyadic captures support two-person interaction and contact-rich manipulation work that single-actor corpora cannot express
• Human interaction / HOI modelling — Two-person motion in one shared coordinate frame, so contact, separation and relative pose are recoverable per frame rather than approximated
• Insurance / Underwriting — Validate rehab claims with objective biomechanical data; assess facility risk via threat intelligence; vet commercial carriers via Spotter sighting bundles
• Physical Therapy — Track patient progress with per-exercise biomechanical reports
• Security & Compliance — Location threat audits, confidence-scored alerting
• Fleet & Logistics / Computer Vision — Commercial-carrier movement corroboration and cargo-fraud investigation from de-identified truck sightings; train truck-recognition detection/OCR models on the geo-stripped CV corpus

Latency

< 500 ms typical for free tools. Cloudflare Workers global edge deployment.

Specifications

• Shared-Frame Multi-Actor Motion Capture — a delivery specification — a vendor-neutral spec for delivering multi-actor mocap without destroying inter-actor geometry, with a pre-purchase verification procedure a buyer can run against any supplier. Draft, published for comment. Canonical URL: https://github.com/RTKMotion/motion-intelligence/blob/main/docs/SHARED_FRAME_MOCAP_SPEC.md

Source

This is a manifest repository for the hosted MCP server. The server implementation runs on Cloudflare Workers; the implementation source is private. For technical questions, partnership, or custom integration: [email protected].

License

The contents of this repository (README, examples, configuration files) are released under the MIT License — see LICENSE.

The data products served by the MCP server are separately licensed via the per-purchase terms documented at api.rtkmotion.io/.well-known/x402 and at rtkmotion.io/legal.