Two endpoints your agent calls before signing on Base: a
multi-aggregator swap router that compares post-fee
output across 0x and Velora on every quote, and a
fail-closed token-contract risk gate that returns
allow / warn / block so your
agent can abstain before signing into a honeypot or sanctioned address.
Built for ElizaOS plugin authors, Coinbase AgentKit integrators, and
custom agent operators on Base.
claude mcp add --transport http --scope user paladin-swap https://swap.paladinfi.com/mcp
Once installed, your agent gets swap_quote,
trust_check_preview (sample fixture; not a real evaluation),
and swap_health as native MCP tools. For production trust
checks, use the paid REST endpoint. Same install works in Cursor or any
MCP-compatible client supporting Streamable-HTTP. Full setup →
Two failure modes routinely cost AI agents real money on Base. The first is signing into a honeypot or sanctioned token contract — an output you can’t exit, an address you shouldn’t touch. The second is routing through one aggregator without comparing — leaving buy-amount on the table because the cheapest route on this minute’s liquidity isn’t the one you hit.
PaladinFi addresses both with a narrow tooling surface, end-to-end vetted, that
you can wire into an agent runtime in one line. Two aggregators audited
end-to-end (per-source 4-byte calldata allowlists, Settler target validation
decoded from inner calldata) is what we’ll claim — not a roster we
haven’t looked at. And on the trust gate: a check that returns allow
when it couldn’t actually run isn’t a gate, so PaladinFi doesn’t.
Multi-aggregator swap router on Base. Returns ready-to-execute calldata for your agent to sign and submit.
# POST /v1/quote
{
"buyAmount": "38421900000000000",
"minBuyAmount": "38229805500000000",
"sellAmount": "100000",
"source": "0x",
"router": "0x0000000000001fF3684f28c67538d4D072C22734",
"calldata": "0x2213bc0b…", // truncated
"trust": { "recommendation": "allow", "version": "1.1", … }
}
Pre-trade composed risk gate for token contracts on Base. Returns
allow / warn / block so your agent can
abstain before signing.
# POST /v1/trust-check ($0.001 USDC via x402)
{
"address": "0x4200000000000000000000000000000000000006",
"chainId": 8453,
"trust": {
"recommendation": "allow",
"risk_score": 0,
"factors": [
{ "source": "ofac", "signal": "not_listed" },
{ "source": "paladin.anomaly", "signal": "address_kind_contract" },
{ "source": "etherscan_source", "signal": "verified" },
{ "source": "goplus", "signal": "trust_list" }
],
"version": "1.1"
}
}treasury.gov/ofac/downloads/sdnlist.xml via systemd timer (operationally may skip during outages — for sub-daily / audit-grade compliance, query Treasury directly). GoPlus and Etherscan run on-call with a short cache. Anomaly heuristics evaluate per request from RPC and Etherscan V2.
signal: "unreachable", real: false; the verdict is derived
from the remaining sources. If all sources are unreachable, the verdict is forced to
warn — we never return allow on a check we couldn’t run.
OFAC SDN hit forces block and is the highest-priority override
(contract details).
Retry-After. The free preview endpoint (/v1/trust-check/preview) caps at 5 concurrent connections per IP.
PaladinFi is built by a small independent infrastructure team focused on the parts of DeFi where protection matters more than novelty: predictable fees, pre-sign safety checks, and tooling that fails safe.
Operated by Malcontent Games LLC (Michigan, USA). Service not offered in the EEA, the United Kingdom, or Switzerland — see the Terms of Service.