@pnllabs/solana-agent-kit-plugin

Trust & risk verdicts for Solana Agent Kit agents — verify BEFORE your agent moves money.

Proof of real PnL, not peak scores. No signup, no API key: free tier per IP/day, then pay-per-call via x402 (USDC on Solana).

🌐 pnllabs.com · 🔌 https://api.pnllabs.com · 🐦 @pnllabs

Why

Every "smart money" score ranks wallets by peak or paper gains. Peaks lie: a wallet can 100x on paper, dump into no liquidity, and walk away red — while its score still says genius. We audited 30 tracked "smart money" wallets: half are genuinely losing money on-chain.

Before your agent copies a wallet, buys a token, or sizes a trade, it should ask four questions — this plugin answers all of them.

Install

npm install @pnllabs/solana-agent-kit-plugin

import { SolanaAgentKit } from "solana-agent-kit";
import PnlLabsPlugin from "@pnllabs/solana-agent-kit-plugin";

const agent = new SolanaAgentKit(wallet, rpcUrl, {}).use(PnlLabsPlugin);

The four checks

| Action | Question | Verdicts | |---|---|---| | CHECK_WALLET_TRUST | Is this wallet worth copying? (real realized PnL) | TRUSTED NEUTRAL UNTRUSTED INSUFFICIENT UNVERIFIABLE | | CHECK_WALLET_FORENSICS | Where did its money come from? (funding origin, freshness, cluster links) | CEX_FUNDED WALLET_FUNDED UNKNOWN_ORIGIN UNVERIFIABLE_ORIGIN | | CHECK_TOKEN_SAFETY | Is this token structurally safe? (concentration, sniper bots, honeypot) | LOW_RISK ELEVATED HIGH_RISK CRITICAL UNKNOWN | | CHECK_TRADE_COST | What will this trade REALLY cost at my size? | ACCEPTABLE_COST ELEVATED_COST HIGH_COST UNTRADEABLE NO_POOL |

Programmatic access without the LLM loop:

const verdict = await agent.methods.checkWalletTrust("57hECC…");
// { verdict: "UNTRUSTED", trust: 15, reasons: [{ code: "NEGATIVE_REALIZED", … }] }

const cost = await agent.methods.checkTradeCost("So1111…112", 2);
// { verdict: "ACCEPTABLE_COST", expected_slippage_pct: 0.4, … }

Honest by design

• Conservative verdicts. When numbers can't be reliably reconstructed the API says UNVERIFIABLE instead of guessing.
• Zero-noise scope. Structural token checks flag structural red flags; behavioral rugs are invisible to all structural checkers — for those, check the wallets involved instead (that's what the wallet checks are for).
• Machine-first schema. Stable verdict enums, reason codes, flags, schema_version on every response.

Payment (x402)

Every check has a daily free tier per IP — try everything without setup. When it runs out, tools return structured x402 payment requirements (payment_required: true) so your agent can pay per call (a few cents USDC on Solana) and retry. Prices & discovery: api.pnllabs.com/.well-known/x402.

Config (optional)

| Key | Default | Purpose | |---|---|---| | PNLLABS_API_BASE | https://api.pnllabs.com | Override API base URL |

Also available as

• Remote MCP (Claude, Grok, ChatGPT, Gemini): https://api.pnllabs.com/mcp
• ElizaOS plugin: @pnllabs/plugin-pnllabs
• Plain HTTP + x402: api.pnllabs.com

MIT © PnL Labs