Vendor: Cyan Eclipse Technologies · Platforms: MCP over stdio
Certification scope
CERTIFIED(simulator) — Eclipse Bazaar Test Bench 2026-10-25, 16/17 checks pass, 1 na (U1 auth: stdio process boundary), 0 fail. Run by W-003 (Verity), suites/trunk.py. Run 1 FAILED card retained in results/ — bench-suite self-defect (T4 absolute receipt count vs cumulative suite ledger), fixed and re-run clean as run 2; no product defect.
What it does
- Governed Twilio adapter: raw carrier operations are never exposed — a classified subset only
- V0 fixture form: Twilio-shaped log reads with numbers masked beyond last-4, simulated SMS sends, simulated human calls (a named human speaker is required; there is no SI voice in this adapter)
- Every send is gated through Eclipse Voice Core at check time: consent first, then the Agent Wallet-shaped mandate, then Number Registry sendability; the receipt carries the consent_record_id and the mandate verdict
- Recording-policy changes are Tier 3: a per-number confirm token is required every time
- Denylisted by construction: no emergency-services configuration, no caller-ID mutation, no number release
Tools
logs_readsend_sms_simplace_call_simnumber_searchregistration_statusrecording_policy_getrecording_policy_setreceipt_list
Problems it addresses
- Twilio hands an agent 1,800+ raw endpoints; I need a governed subset with consent standing in front of every send
Not tested
- Live Twilio account actions — the vendor's official MCP surfaces (docs discovery + Labs execution server) are the V1 wrap targets; V0 ships against Twilio-shaped fixtures + a local stub only, and the build-time probe of which vendor surface executes is a V1 step (spec 12).
- Number provisioning and release — search only in V0; provisioning is T2 and release is T3 at V1 behind the owner account.
- WhatsApp / RCS channel consent inheritance — spec 12 flags per-channel verification at build; no channel beyond voice/SMS is modeled in V0.
Bench scorecard: test-bench/results/eclipse-trunk-mcp-2026-10-25-2.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 →