AI agents invoke emergency_stop to trigger actions in CHAI pentest tool. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
While not destructive of data, an emergency stop is an Execute-class tool because it triggers an external operation (terminating processes) whose consequences depend on the argument (session ID). The high severity reflects that stopping a penetration test session mid-operation could disrupt critical security testing, strand incomplete scans, or impact system state.
From the tool's definition emergency_stop stops all processes for a session; this is a forceful operational action that terminates running operations, which constitutes execution of a control operation whose effects depend on which session is targeted.
Documented attack patterns abuse exactly the kind of access emergency_stop gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and CHAI pentest tool, and nothing reaches the server without passing your rules. This is the rule we recommend for emergency_stop:
{
"version": "1",
"default": "deny",
"tools": {
"emergency_stop": {
"limits": [
{
"counter": "emergency_stop_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}emergency_stop stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Emergency stop all processes for a session. It is categorised as a Execute tool in the CHAI pentest tool MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the CHAI pentest tool MCP server in PolicyLayer and add a rule for emergency_stop: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches CHAI pentest tool. Nothing to install.
emergency_stop is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the emergency_stop rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for emergency_stop. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
emergency_stop is provided by the CHAI pentest tool MCP server (nihar-sarkar/chai). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from CHAI pentest tool, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
16 CHAI pentest tool tools catalogued and risk-classified — across an index of 43,000+ MCP servers.