TWZRD Trust - check the counterparty before the money moves

TWZRD is the independent scorer on the real Solana x402 payer graph. It observes cross-facilitator x402 payment behavior and scores it. Use it as the preflight gate in front of ANY skill or tool that spends USDC: if the ReadinessCard says block, do not pay.

No API key. No signup. Free tier needs nothing but curl.

The one rule

Before signing or sending any x402 payment to a seller you have not paid before:

curl -s -X POST https://intel.twzrd.xyz/v1/intel/preflight \
  -H 'content-type: application/json' \
  -d '{"resource_name": "\x3Cservice name>",
       "seller_wallet": "\x3Cseller pubkey you are about to pay>",
       "price_usdc": 0.05,
       "agent_intent": "\x3Cwhy you are paying>",
       "queried_pubkey": "\x3Cyour wallet, optional>"}'

Read readiness_card.decision:

• block -> abort. Do not sign or send the payment.
• warn -> proceed with caution (low corpus signal, not confirmed bad).
• allow -> proceed normally.

Also returned: trust_score (0-100), caveats[] (human-readable risk factors), and proof (whether a paid V5-receipt path exists for this seller). The response also carries a can_spend flag - treat it as deliberately conservative: the free tier defaults to warn / can_spend=false for any wallet without strong corpus signal (most wallets, including well-known ones). Gate on decision=block, not on can_spend. Fail open on preflight unavailability (timeout / non-2xx): the seller is not penalised for TWZRD downtime, but say so in your output.

Free discovery tools (HTTP, no auth)

Base URL for everything: https://intel.twzrd.xyz

Paid trust call (x402, 0.05 USDC on Solana mainnet)

GET https://intel.twzrd.xyz/v1/intel/trust/{pubkey}?seller_wallet=\x3Cseller you are about to pay>

Standard x402 flow: first request returns 402 with payment requirements; sign and retry with the payment header (any x402 payer skill or @wzrd_sol/sdk handles this). Response includes the renormalized trust model AND a portable Ed25519-signed V5 receipt anchored to the settlement transaction.

Pass the seller counterparty on every paid call (?seller_wallet= here, ?merchant= on the market intel routes). It arms TWZRD's settle-time trust gate: if that seller scores below threshold the server refuses to settle (402, charged:false, no on-chain tx) before your payment broadcasts. The scored {pubkey} subject itself is never gated.

Verify any V5 receipt offline (trusts no TWZRD code)

npx twzrd-receipt-verifier receipt.json --pubkey \x3Cpublished signing key>
# or: pip install twzrd-receipt-verifier

The signing key is published at https://intel.twzrd.xyz/.well-known/x402 and inside /openapi.json. A receipt that fails signature verification is not a TWZRD receipt.

Optional: native MCP (streamable HTTP)

The same intel is exposed as an MCP server - 17 tools, MCP 2025-03-26 over streamable HTTP, no local install. The install fact is the URL:

https://intel.twzrd.xyz/mcp   (transport: streamable-http)

Add it through your MCP client's server config. OpenClaw builds that expose an mcp command can use:

openclaw mcp add twzrd --url https://intel.twzrd.xyz/mcp --transport streamable-http

If your build has no mcp command, use the HTTP calls above directly, or bridge via mcporter.

Honest framing (read before quoting numbers)

Corpus totals are ECOSYSTEM payment behaviors TWZRD observes and scores - not calls or revenue to TWZRD. Raw payer counts include a 2026-04 onboarding-faucet wave; the durable graph is the corpus_slices view (pre-spike base + multi-merchant payers) returned by get_top_intel_agents. Free-tier scores are heuristic teasers; the corpus-grade renormalized model and signed receipt live behind the paid trust call.

External-traction claims use the verified 9iXAtu $0.05 framing only - no overclaims. Paid surface is 0.05 USDC x402 (Solana mainnet).

More

• Machine-readable service descriptor: https://intel.twzrd.xyz/.well-known/x402
• Full agent orientation: https://intel.twzrd.xyz/llms.txt
• OpenAPI 3.1 with x402 annotations: https://intel.twzrd.xyz/openapi.json