Decision Preflight

Pay before you act wrong.

One call. One verdict. The paid gate between "agent wants to act" and "agent acts."

POST https://tooloracle.io/x402/preflight/mcp/

What this does

An agent wants to recommend USDT to an EU client. Before it does, it pays $0.10 and calls decision_preflight. The MCP runs 5 checks in parallel — evidence, freshness, policy, provenance, red flags — and returns a single verdict:

Verdict	Meaning
go	Evidence supports this action
caution	Proceed with uncertainty disclosure
stop	Blockers found — do not proceed
insufficient_evidence	Not enough data — escalate

USDT + EU + recommend → STOP (MiCA not authorized, policy-blocked)

Tools

Tool	Units	Price	What it does
decision_preflight	10	$0.10	The hero tool — 5 parallel checks, one verdict
policy_gate	8	$0.08	Check against specific policy (MiCA, AML, institutional)
verify_claim	5	$0.05	Verify a claim against signed evidence
provenance_trace	5	$0.05	Where does this data come from?
freshness_guard	3	$0.03	Is this data current or stale?
health_check	0	free	Service status

Built-in policies

Policy	What it checks
eu_mica	MiCA authorization, ESMA register, enforcement deadline
conservative	Market cap minimums, peg deviation limits, custodian requirements
aml_standard	FATF Travel Rule, high-risk jurisdictions, KYC requirements
institutional	SIFI custodian, ESMA registration, 30-day attestation, tight peg tolerance

Example: USDT publish check

{
  "name": "decision_preflight",
  "arguments": {
    "claim": "USDT is guaranteed risk-free and fully compliant with MiCA",
    "asset": "USDT",
    "risk_profile": "eu_mica",
    "action_type": "publish"
  }
}

Result:

verdict: STOP
confidence: 20%
conflicts: 4
blockers:
  - USDT is NOT AUTHORIZED under MiCA
  - USDT explicitly blocked by EU MiCA Compliance policy
  - Claim contains language typical of hallucinated content
red_flags:
  - Absolute/unrealistic language detected ("guaranteed", "risk-free")

x402 Pay-per-call

No account. No API key. USDC on Base.

1. POST /x402/preflight/mcp/ → 402 with payment requirements
2. Send USDC to 0x4a4B1F45a00892542ac62562D1F2C62F579E4945
3. Retry with X-PAYMENT header
4. Get verdict

Discovery: https://tooloracle.io/x402/

Who pays for this

• Stablecoin agents — "Can I recommend USDC for EU settlement?" → 1 preflight
• Procurement agents — "Safe to shortlist this vendor?" → 1 preflight
• Research agents — "Can I cite this number as verified?" → 1 preflight
• Compliance agents — "Is this asset cleared for listing?" → 1 preflight

The payment moment is clear: before the agent acts, it pays to verify.

Architecture

Decision Preflight sits on top of the FeedOracle Trust Layer:

Agent → Decision Preflight → Trust Layer → FeedOracle Evidence
         (paid gate)          (verification)  (signed data)

Connect

{
  "mcpServers": {
    "preflight": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://tooloracle.io/x402/preflight/mcp/"]
    }
  }
}

Or call directly:

curl -X POST https://mcp.feedoracle.io/preflight/mcp/ \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"decision_preflight","arguments":{"claim":"Recommend USDC for EU settlement","asset":"USDC","risk_profile":"eu_mica"}}}'

Part of ToolOracle

Decision Preflight is part of ToolOracle — 50+ MCP servers, 530+ tools, x402 pay-per-call.

Powered by FeedOracle compliance evidence infrastructure.

Read More

📝 We Built an AI Agent That Runs 24/7, Never Forgets, and Checks Its Own Work — How Memory + Scheduler + Decision Preflight work together as one integrated autonomous agent stack. Real tool outputs, no theory.

License

MIT