The Trust Layer
for AI Agents
Before your agent acts, it checks TrustGrade. Bayesian trust scores for DeFi protocols, stablecoins, and smart contracts — built for the age of autonomous transactions.
Three steps to safe autonomous transactions
Agent receives task
Your AI agent gets an instruction: swap tokens, provide liquidity, interact with a protocol.
Agent calls TrustGrade
Before executing, the agent queries TrustGrade via API or MCP to check the trust score of the target entity.
Agent executes or aborts
If the trust score passes the threshold, the agent proceeds. If not, it aborts and logs the reason.
Integrate in under 60 seconds
One API call. One MCP tool. Zero excuses for untrusted agent transactions.
from trustgrade import TrustGrade
client = TrustGrade(api_key="tg_live_xxx")
# Check a DeFi protocol before your agent interacts
result = client.score("defi_protocol", "Uniswap")
print(f"Grade: {result.grade} Score: {result.score}/100")
print(f"Confidence: {result.confidence:.0%}")
print(result.summary)
# Agent decision logic
if result.score >= 80:
agent.execute_transaction()
else:
agent.abort("TrustGrade score below threshold")import { TrustGrade } from '@trustgrade/sdk';
const tg = new TrustGrade({ apiKey: process.env.TRUSTGRADE_KEY });
// Pre-transaction trust check
const score = await tg.score('stablecoin', 'USDC');
if (score.grade === 'A+' || score.grade === 'A') {
await agent.swap(tokenIn, tokenOut, amount);
console.log('\u2705 Transaction executed — trust verified');
} else {
console.log(`\U0001F6AB Aborted: ${score.grade} grade too low`);
console.log(score.summary);
}Built into the tools your agents already use
TrustGrade ships as an MCP server. Your agents can check trust scores directly from Claude Desktop, Cursor, or any MCP-compatible client. No wrapper code needed.
- ✅ 7 MCP tools — check_trust, verify_transaction, verify_agent, and more
- ✅ One-line install —
npx @trustgrade/mcp-server - ✅ Works with — Claude Desktop, Cursor, Windsurf, and any MCP client
{
"mcpServers": {
"trustgrade": {
"command": "npx",
"args": ["-y", "@trustgrade/mcp-server"],
"env": {
"TRUSTGRADE_API": "https://api.trustgrade.ai",
"TRUSTGRADE_API_KEY": "tg_live_xxx"
}
}
}
}Autonomous transactions require autonomous verification
Speed without recklessness
Agents act in milliseconds. TrustGrade returns cached scores in under 50ms — fast enough for real-time decisions.
Mathematically defensible
Every score is backed by Wilson confidence intervals and Bayesian posteriors. Not vibes — statistics.
Pre-transaction verification
Block bad transactions before they hit the chain. Agents check trust, then act. Not the other way around.
Institutional grade
Built for protocols, market makers, and custodians who need audit-ready trust signals for compliance.
Universal integration
REST API, MCP server, SDK embeds. Drop trust scoring into any agent framework — Eliza, Autogen, CrewAI, or custom.
Open registry
45+ entities scored and growing. Anyone can query, verify, or embed trust badges for free.
Start scoring in minutes
Free tier: 100 API calls/day. No credit card. No lock-in.