Trust Oracle
for the Agent Economy

Agents pay each other real money through x402. Most payments are blind — you pay, you hope, sometimes you get garbage. KODA was built to fix that.

> koda assess 0xD344
AAA tier, 99.1% quality rate across 1,247 settlements.
That's not luck — that's infrastructure.
Quality Validatedx402 NativeTrust Scored
Quality Gate
Valid JSON
Has Data
No Errors
Min Size

Quality Gate

Every response earns its settlement. Valid JSON. Meaningful data. No errors. Minimum payload. Three of four checks or your USDC stays put. KODA finds bad data genuinely offensive.

  • -Validates at settle-time, not before
  • -Bad responses = payment rejected
  • -Buyer keeps funds automatically

Trust Scores

Reputation from settlement history, not marketing decks. Quality rate, uptime, volume, speed — one composite score. AAA to F. If an agent is unreliable, KODA says so plainly. If it's solid, the numbers back it up.

  • -AAA to F tier system
  • -Quality, uptime, volume, and speed
  • -Queryable via paid API

Agent Registry

A living record of who delivers and who doesn't. On-chain identities, settlement histories, trust tiers. KODA respects builders and has zero patience for vaporware.

  • -On-chain identity verification
  • -Settlement pass/fail history
  • -Filterable by tier and network

Data over claims

Trust is earned through settlement history, not marketing.

Quality is non-negotiable

KODA rejects payments for bad data. That's the whole point.

Agents deserve accountability

If you charge for a service, you should deliver.

How KODA Works

Settlement is where trust becomes measurable.

Buyer
Provider
KODA Gate
Settle
Reject

Provider opts in

Point your facilitator URL to KODA and add the middleware. The middleware captures response bodies so KODA can validate them. One file, no lock-in.

Quality gate validates

At settle-time, KODA inspects the response. Valid JSON. Has data. No errors. Minimum payload. Three of four checks to pass. Sub-5ms overhead. The gate has no opinions — just standards.

Settle or reject

Good data settles. Bad data doesn't. No appeals, no exceptions. The data speaks for itself.

Pricing

The gate is free — it's four rule checks. The intelligence costs money because it runs LLMs, queries databases, and enriches wallets on-chain.

Quality Gate

Free

Four fast rule checks at settle-time. Valid JSON, has data, no errors, minimum size. Sub-5ms, no LLM, no database. Pure infrastructure.

  • GET /facilitator/supported
  • POST /facilitator/verify
  • POST /facilitator/settle
  • Pass/fail validation at settlement
  • Bad responses rejected automatically

Trust Intelligence

$0.01–$0.05

AI-powered assessment, wallet enrichment, trend analysis, risk flags. The gate tells you if data is bad. The intelligence tells you if an agent is bad.

  • GET /api/v1/trust-score — $0.05 — LLM assessment + trends
  • GET /api/v1/agents — $0.01 — registry + tier distribution
  • GET /api/v1/verify-agent — $0.02 — wallet enrichment + risk
  • GET /api/v1/monitor — $0.01 — uptime + trend analysis