The kill switch
AI agents need.
Enterprise-grade safety infrastructure for autonomous AI. Real-time monitoring, instant termination, and cryptographic identity—all in under 100ms.
10
Security Modules
<100ms
Kill Latency
6,149
Lines of Code
30s
SPIFFE Revocation
One decorator.
Complete protection.
Add enterprise-grade safety to any AI agent with a single line of code. Works with OpenAI, Anthropic, LangChain, and any custom implementation.
from killswitch import fence
# That's it. Your agent is now protected.
@fence.protect(agent_id="prod-agent-001")
def my_ai_agent(task):
result = llm.complete(task)
return result
Defense in depth
10 security modules working together to ensure your AI agents stay within bounds
Runtime Fence
Monitor every action in real-time with configurable boundaries and instant alerts
Instant Kill Switch
One-click termination with SIGTERM → SIGKILL escalation for rogue agents
Audit Logging
Complete audit trail with cryptographic verification for compliance
Sub-Second Response
Kill signals processed in under 100ms when safety matters most
SPIFFE Identity
Cryptographic workload identity with automatic certificate rotation
Token Governance
Decentralized oversight with $KILLSWITCH token voting rights
Stake in AI safety
Token holders govern protocol upgrades, earn subscription discounts, and participate in the decentralized oversight network.
Contract Address
56o8um92XU8QMr1FsSj4nkExEkgKe56PBTAMqCAzmoonSolana • SPL Token
Ready to secure your AI agents?
Get started in minutes. Free tier for personal projects, enterprise plans for production workloads.