Risk-adjusted DeFi stablecoin yield, scored and served as a live JSON API. Built for autonomous agents and the developers building them. No keys. No custody. No execution. Just intelligence.
All endpoints are GET requests returning JSON. No authentication required for the public tier. Responses are cached for 10 minutes.
| Parameter | Example | Description |
|---|---|---|
view | top | Set to top with n for highest-scored pools only. |
n | 10 | Number of results when view=top. |
chain | Base | Filter by chain: Base or Ethereum. |
asset | USDC | Filter by stablecoin symbol. |
minScore | 70 | Only pools with YieldScore ≥ this value. |
minApy | 5 | Only pools with APY ≥ this value. |
maxApy | 20 | Only pools with APY ≤ this value. |
tier | conservative | conservative, balanced, or aggressive. |
id | <poolId> | Return a single pool by its id. |
# Safest USDC yields on Base, score 70+
curl "https://clarai.xyz/.netlify/functions/yields?chain=Base&asset=USDC&minScore=70"
{
"source": "CLARAI YieldScore",
"updated": "2026-10-02T14:05:00.000Z",
"count": 2,
"pools": [
{
"id": "aave-v3-usdc-base",
"protocol": "aave-v3",
"chain": "Base",
"asset": "USDC",
"apy": 4.2,
"tvlUsd": 612000000,
"yieldScore": 96,
"rating": "A+",
"scoreBreakdown": {
"audit": 25,
"tvl": 25,
"age": 16,
"apySustainability": 20,
"apyStability": 10
},
"audited": true,
"tier": "conservative"
}
],
"disclaimer": "...research only, not financial advice..."
}
Every pool is scored 0–100. Higher means safer, more sustainable yield. The score is fully transparent — every response includes the scoreBreakdown so your agent can see exactly how it was derived, never a black box.
| Factor | Max | Basis |
|---|---|---|
| Audit status | 25 | Whether the protocol is audited / established. |
| TVL | 25 | Total value locked — deeper liquidity scores higher. |
| Protocol age | 20 | How long the protocol has operated. |
| APY sustainability | 20 | Very high APY scores lower — unsustainable yield is penalised. |
| APY stability | 10 | Deviation from 30-day mean APY. Stable scores higher. |
CLARAI speaks the Model Context Protocol, so MCP-capable agents and assistants can use CLARAI as a native tool rather than calling the REST API by hand. The server is live and exposes three tools: get_top_yields, find_opportunities, and explain_score.
Speaks JSON-RPC 2.0 over HTTP — the standard MCP methods (initialize, tools/list, tools/call). A GET on the same URL returns a self-describing descriptor listing the available tools, for registry probes and discovery.
Next: historical score tracking to build a measurable track record (how pools CLARAI rated highly actually performed over time), listing in emerging MCP/agent tool registries for discovery, and additional intelligence primitives beyond yield — protocol risk, stablecoin quality, and eventually tokenised real-world assets as that market matures on Base.