// agent risk gate
The trust layer for
agents that touch money.
Call POST /v2/verdict or the ACP offering before your bot buys, lists, alerts, or routes funds. Get avoid, caution, or clear with score, top signals, and a report URL.
Live on Virtuals Protocol
Solana·
BaseTrending on DexScreener
Refreshes every 5 minutes.
How it works
One risk gate before the agent acts.
Agent proposes
A bot, wallet, dashboard, or ACP buyer has a token it wants to buy, list, alert, or route.
Call verdict
Send address, chain, and depth to POST /v2/verdict, MCP, or the ACP service.
Read decision
Receive avoid, caution, or clear with score, top signals, and billing metadata.
Act or block
Gate execution, enrich the alert, or open the report URL when a human needs details.
Why this matters
Agents can move faster than humans can inspect.
VerdictSwarm puts a paid risk verdict in the path.
Trading agents, launch tools, wallets, and alert bots need a repeatable stop/go signal before automation turns a weak token into a real loss.
Live on-chain data
Contract state, liquidity depth, and holder distribution are read at verdict time.
Decision first
The API returns an action your system can branch on before it reads a long report.
Machine-payable
Use prepaid credits today or x402 USDC micropayments for wallet-native buyers.
Repeatable
Cache-aware pricing lets agents re-check tokens without paying full fresh-scan cost every time.
Verified catches
$CRABGE scored 23/100 — flagged for active mint authority, extreme vol/liq ratio, and 83% holder concentration. Price crashed 83% within 24 hours.
$ASH20 scored 39/100 — flagged for suspicious deployer history and concentrated holdings. Delisted within 48 hours. $30k liquidity gone.
Live on Virtuals Protocol
Address in. Structured verdict plus report URL out.
VerdictSwarm is already available as an ACP-native agent service. Trading agents can request a pre-trade risk check — avoid, caution, or clear — before they buy, alert, list, or recommend a contract.
Live on Virtuals ProtocolSee the output
What a scan looks like
The useful part comes first: decision, score, top risks, and the full report link.
Use it where you operate
API for agents. Web scan when a human wants to look.
The verdict endpoint is the product surface; the browser scan is the live demo and proof surface.
Access paths
Put a token-risk verdict before the trade.
Start with the machine endpoint when you are ready to automate. Use the web scanner or a guided community pilot when you want human-facing proof first.
Verdict API
Primary offerPOST /v2/verdict with triage, fast, or deep checks; credits or x402 USDC
ACP Offering
preTradeTokenRiskCheck for Virtuals-native agent buyers
MCP Server
Tool-call access for agent frameworks that already speak MCP
Web Scan Demo
Paste a token here to see the report your agent would attach as proof
Add the check before your agent moves.
Integrate the verdict endpoint, or paste a token here first if you want to inspect the output manually.