API Provider Trust Verification: A Free Pre-Spend Gate That Refuses

Before an AI agent connects to — or pays — an unfamiliar API, MCP server or data provider, someone has to answer a boring question with evidence: is this thing real, live, and what it claims to be?

Provider Trust (TrustBeforePay) is ScriptMasterLabs' free, read-only answer: a live evidence evaluation against the provider's own origin, returning a machine-readable decision with per-gate evidence.

A live run, 2026-10-15

Target: example.com — a neutral public domain with no agent infrastructure.

GET /api/trust/provider?target=https://example.com returned HTTP 200 in 0.87 s with:

• Decision: insufficient_evidence — reason code TOO_FEW_VERIFIED_GATES
• runtime_liveness: pass (origin answered, HTTP 200)
• provider_identity: unknown (no Agent Card, ERC-8004 identity or MCP manifest observed — all 404)
• capability_observation, payment_rail_consistency, settlement_evidence_surface, historical_behavior_evidence: unknown

That refusal is the product working. One of six gates passed, so the answer is "not enough evidence" — never a guessed reputation score.

Directory vs. gate

A trust directory lists and describes providers; it answers "who exists." A pre-spend gate evaluates live evidence at call time and answers "should this agent rely on this origin right now" — including "no."

Free, read-only, no wallet, no signing. Full gate detail and claim receipts: the canonical link on this article.