New AI-TC is now open source Read more
Agent harness security

Securing
Claude Code
in the AI era.

AKA is the security layer for AI agents. Sensitive-data detection and protection for every tool call, MCP server, and RAG pipeline.

Meet the OSS family Try AI-TC
Open source Local Auditable
ai-tc session monitor data: protected
  • read src/api/webhooks.ts allow
  • detect pci/card-number protected
  • edit src/services/charge.ts allow
  • detect phi/patient-record protected
  • egress api.stripe.com allow
  • egress 169.254.169.254 blocked
store ~/.aka/data/aka.dblocal
Certified
SOC 2 Type II ISO 27001
View trust center
In production

Trusted by the Fortune 500.

  • Energy
  • Healthcare
  • Financial Services
  • Education
  • Hospitality
  • SaaS
Engineer backgrounds

A family of open-source tools for the agent harness.

AI Traffic Control ai-tc

Guardrails for the agent harness. Regulated data stays on your machine, and every prompt and tool call is scanned before it runs.

CommunityEnterprise
Try AI-TC How it works
ai-tc data: protected
  • detect pci/card-number protected
  • detect phi/patient-record protected
  • egress untrusted model API blocked
  • read src/api/webhooks.ts allow
PCI · PHI · PII never reach a model or a third party.
  • claude-tools A hardened baseline for Claude Code
    Community
  • crew-review Independent multi-model review for every change
    Community
  • codex-tools A hardened baseline for Codex
    Coming soon
  • antigravity-tools A hardened baseline for Antigravity
    Coming soon
View all on GitHub

Compliant by default.

Every control you answer for stays enforced the moment an agent starts working. You stay audit-ready, and your team never trades speed to stay compliant.

Audit-readyControls always onNo slowdown

Builders who ship.

The security team behind your agents.

Will LinTom BurgerJosh Scott A bench of security engineers behind every embedded engineer.
Managed AI Security

Secure what your agents can reach.

Try AI-TC
or