Stops runaway AI agent loops and token burn. Checks each tool call before it runs. No signup.
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.
Run a live HTTP audit against a public domain and return score, grade and report URL.
Target domain, e.g. cursor.com
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 agentguard https://agentguard.audiscape.com/mcp
Run a live HTTP audit against a public domain and return score, grade and report URL.
| Argument | Type | Description |
|---|---|---|
| mode | string | |
| domain* | string | Target domain, e.g. cursor.com |
{
"type": "object",
"required": [
"domain"
],
"properties": {
"mode": {
"enum": [
"citerank",
"checkoutguard",
"agentguard",
"signalstrike"
],
"type": "string"
},
"domain": {
"type": "string",
"description": "Target domain, e.g. cursor.com"
}
}
}Return a Stripe checkout link for the remediation pack for a domain.
| Argument | Type | Description |
|---|---|---|
| mode | string | |
| domain* | string |
{
"type": "object",
"required": [
"domain"
],
"properties": {
"mode": {
"enum": [
"citerank",
"checkoutguard",
"agentguard",
"signalstrike"
],
"type": "string"
},
"domain": {
"type": "string"
}
}
}Return the remediation preview and scorecard URL for a domain.
| Argument | Type | Description |
|---|---|---|
| domain* | string |
{
"type": "object",
"required": [
"domain"
],
"properties": {
"domain": {
"type": "string"
}
}
}Replay an agent's tool-call log through the AgentGuard loop and budget guard and decide whether the next call should run. Catches identical and near-repeat loops (same file, URL or command), re-reads of finished items, failing-tool retry streaks and USD or token caps before spend.
| Argument | Type | Description |
|---|---|---|
| next* | object | |
| calls* | array | |
| config | object | Optional agentguard.json contents |
{
"type": "object",
"required": [
"calls",
"next"
],
"properties": {
"next": {
"type": "object",
"required": [
"tool"
],
"properties": {
"tool": {
"type": "string"
},
"arguments": {}
}
},
"calls": {
"type": "array",
"items": {
"type": "object",
"required": [
"tool"
],
"properties": {
"tool": {
"type": "string"
},
"tokens": {
"type": "number",
"minimum": 0
},
"costUsd": {
"type": "number",
"minimum": 0
},
"isError": {
"type": "boolean"
},
"readOnly": {
"type": "boolean",
"description": "Call had no side effects; lets the opt-in completed-read rule apply"
},
"arguments": {},
"errorText": {
"type": "string",
"description": "Error message when isError; permanent errors (missing binary, 4xx, quota, permission) halt after 1 retry"
},
"resultHash": {
"type": "string",
"description": "Stable hash of the tool result; repeats with the same hash count as no progress"
},
"totalTokens": {
"type": "number",
"minimum": 0,
"description": "Running session tokens so far"
},
"totalCostUsd": {
"type": "number",
"minimum": 0,
"description": "Running session spend so far; spend becomes max(spend, total)"
}
}
},
"maxItems": 20000
},
"config": {
"type": "object",
"description": "Optional agentguard.json contents"
}
}
}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 Nov 1, 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.