Pre-sign Solana tx guard for agents: ALLOW/WARN/BLOCK verdict, x402 USDC pay-per-call.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Use this before an agent or wallet signs a Solana transaction. Decodes the unsigned or signed transaction and returns NO_MATCH (no known-bad pattern matched; not a safety guarantee), WARN, or BLOCK, with plain-language reasons and a signed receipt. Pass an x402 quote to also check a payment transaction against it (overpay, approve-instead-of-transfer, unchecked Transfer, wrong payee). Errors return ERROR and are never charged. Price $0.009 USDC (x402, Solana).
Base64 wire transaction (also accepts an RPC getTransaction result).
Optional spend caps you declare; any violation returns BLOCK. Keys: max_amount (int base units per outflow), max_amounts ({mint|'SOL': int}), max_total (int per asset), allowed_payees ([pubkey]; owners' ATAs match), allowed_assets ([mint|'SOL']), allow_approvals (bool, default false).
x402 quote the tx pays: {asset, amount, pay_to | pay_to_token_account, decimals?, token_program?}.
Escalate HIGH-risk token approvals (unlimited or third-party delegate) from WARN to BLOCK. Default false.
Your own pubkey; only this key is treated as 'you'.
Simulate against Solana mainnet (Helius RPC, sigVerify off, recent blockhash replaced, never submitted). Default true; false = static checks only. If simulation fails the verdict is at least WARN, never NO_MATCH.
Hard per-payment cap enforced even if the quote asks for more.
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
claude mcp add --transport http falcon-shield https://falcon-shield.falcon-sol.workers.dev/mcp
Use this before an agent or wallet signs a Solana transaction. Decodes the unsigned or signed transaction and returns NO_MATCH (no known-bad pattern matched; not a safety guarantee), WARN, or BLOCK, with plain-language reasons and a signed receipt. Pass an x402 quote to also check a payment transaction against it (overpay, approve-instead-of-transfer, unchecked Transfer, wrong payee). Errors return ERROR and are never charged. Price $0.009 USDC (x402, Solana).
| Argument | Type | Description |
|---|---|---|
| tx* | string | Base64 wire transaction (also accepts an RPC getTransaction result). |
| caps | object | Optional spend caps you declare; any violation returns BLOCK. Keys: max_amount (int base units per outflow), max_amounts ({mint|'SOL': int}), max_total (int per asset), allowed_payees ([pubkey]; owners' ATAs match), allowed_assets ([mint|'SOL']), allow_approvals (bool, default false). |
| quote | object | x402 quote the tx pays: {asset, amount, pay_to | pay_to_token_account, decimals?, token_program?}. |
| strict | boolean | Escalate HIGH-risk token approvals (unlimited or third-party delegate) from WARN to BLOCK. Default false. |
| wallet | string | Your own pubkey; only this key is treated as 'you'. |
| simulate | boolean | Simulate against Solana mainnet (Helius RPC, sigVerify off, recent blockhash replaced, never submitted). Default true; false = static checks only. If simulation fails the verdict is at least WARN, never NO_MATCH. |
| cap_base_units | integer | Hard per-payment cap enforced even if the quote asks for more. |
{
"type": "object",
"required": [
"tx"
],
"properties": {
"tx": {
"type": "string",
"description": "Base64 wire transaction (also accepts an RPC getTransaction result)."
},
"caps": {
"type": "object",
"description": "Optional spend caps you declare; any violation returns BLOCK. Keys: max_amount (int base units per outflow), max_amounts ({mint|'SOL': int}), max_total (int per asset), allowed_payees ([pubkey]; owners' ATAs match), allowed_assets ([mint|'SOL']), allow_approvals (bool, default false)."
},
"quote": {
"type": "object",
"description": "x402 quote the tx pays: {asset, amount, pay_to | pay_to_token_account, decimals?, token_program?}."
},
"strict": {
"type": "boolean",
"description": "Escalate HIGH-risk token approvals (unlimited or third-party delegate) from WARN to BLOCK. Default false."
},
"wallet": {
"type": "string",
"description": "Your own pubkey; only this key is treated as 'you'."
},
"simulate": {
"type": "boolean",
"description": "Simulate against Solana mainnet (Helius RPC, sigVerify off, recent blockhash replaced, never submitted). Default true; false = static checks only. If simulation fails the verdict is at least WARN, never NO_MATCH."
},
"cap_base_units": {
"type": "integer",
"description": "Hard per-payment cap enforced even if the quote asks for more."
}
}
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Oct 30, 2026.
No readOnlyHint / destructiveHint annotations, so clients can't tell a read from a write without asking.
Fix: Add annotations to each tool, e.g. { readOnlyHint: true } for tools that only read.