Vendor: Cyan Eclipse Technologies · Platforms: MCP over stdio (HTTP transport on roadmap)
Certification scope
CERTIFIED(simulator) — Eclipse Bazaar Test Bench 2026-10-25, 14/15 checks pass, 1 na (U1 auth: stdio process boundary). Run by W-003 (Verity), suites/environment.py.
What it does
- Device Registry: unenrolled devices are invisible to every tool, even when the hub can see them
- Safety Policy ladder: Tier 0 sense, Tier 1 benign/reversible, Tier 2 consequential, Tier 3 safety-critical
- Spec floors per capability class: overrides may only raise tiers; a config that lowers one is rejected at load
- State-confirmed execution only: a command the device does not confirm earns no receipt, only an alert record
- Tier 1 execution in v0.1; Tier 2/3 evaluate and deny by phase; kill switch freezes actuation while sensing stays live
Tools
registry_liststate_snapshotpolicy_evaluatecommand_executereceipt_list
Problems it addresses
- My smart-home hub hands an agent every control by default — I need scoped authority
- A lock or a heating appliance must never be one config edit away from an agent
- Receipts should assert only effects the device itself confirmed
Not tested
- A real Home Assistant instance — v0.1's hub is a local HA-shaped stub (stdlib http, fixtures). The wrapper's client speaks the /api/states + /api/services shape, but parity with real HA behavior (auth, service payloads, state semantics) is [verify] and untested; connecting the owner's instance is a Wave-2 owner-gated step.
- Tier 2 / Tier 3 EXECUTION — by phase design v0.1 evaluates both and denies by phase; the Tier 2 receipt + owner-notification wiring and any Tier 3 execution path are unbuilt, so there is nothing to test yet.
- Vehicle rails (Smartcar, Tesla Fleet API) — designed in the Procure concept doc's Environment module; no adapter exists in v0.1.
- Real-home scale and latency — fixture hub carries 7 entities; nothing about a real home's entity count, hub latency, or state churn has been measured.
Bench scorecard: test-bench/results/eclipse-environment-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 →