Vendor: Cyan Eclipse Technologies · Platforms: MCP over stdio (HTTP transport on roadmap)
Certification scope
CERTIFIED(simulator) — Eclipse Bazaar Test Bench 2026-10-25, 16/17 checks pass, 1 na (U1 auth: stdio process boundary). Run by W-003 (Verity), suites/zenith.py.
What it does
- FTMS codec (treadmill, indoor bike, rower, heart rate, control point) proven against hand-computed byte fixtures in both directions
- Equipment Registry: unenrolled equipment invisible to every tool
- Session state machine (idle, active, paused, ended); ending writes the workout record as a Data Rights Ledger exhaust record
- Tiered session control: treadmill speed is Tier 3 behind a confirmation token, with per-command delta bounds and device/session ceilings
- Dead-man rule enforced in the equipment layer: lapse the control heartbeat and the simulator stops the belt itself, receipted as a safety action
- Tier floors validated at load: a config declaring speed below Tier 3 exits before serving
Tools
registry_listtelemetry_readsession_startsession_controlsession_endworkout_listreceipt_list
Problems it addresses
- An agent coaching workouts must not spin up a treadmill unchecked
- If the controller goes silent mid-run, the belt must stop anyway
- The training log should be the session receipts themselves
Not tested
- Real FTMS hardware of any model — none exists on this VM. The codec follows the published FTMS characteristic layouts, but per-model variance (optional-field population, flag discipline, control-point parameter acceptance, rower pace resolution) is [verify] against each machine's advertised characteristics at enrollment and is untested here; the registry's characteristics lists describe the SIMULATOR's advertised set.
- Real Bluetooth transport (BLE pairing, notification cadence ~250 ms, link loss behavior) — the bridge-to-simulator link is a local Unix socket; the dead-man rule is proven on a simulated clock against the simulator's enforcement, not against a real belt, a real link, or real time.
- Physiological ceiling behavior with a real athlete (HR-signal loss handling, injury/medical flags, Coach adaptation) — designed in the concept doc; the Coach engine and Health Fusion are unbuilt, so there is nothing to test yet.
- Multi-user / commercial-gym identity — deferred by the concept doc; v0.1 is home-first, single-athlete fixtures.
Bench scorecard: test-bench/results/eclipse-zenith-mcp-2026-10-25.json
Evaluation-intake listing
This listing records the tool's place in the Bazaar's evaluation
and monitoring pipeline, with the evidence gathered so far printed above — verdict, scope,
and not-tested items included. It is not sold on this site, and its presence in the catalog
is not a claim beyond that evidence. The service behind every verdict is the bench:
what a bench run buys →