On this page
Verosek builds one layer
between your AI agents and the models and tools they call.
Verosek is an AI + MCP Gateway for AI agent security. It routes model calls and, for the tools it runs, reads what each call does, down to the SQL statement, then decides it against the agent key's rules before it runs. Scanning and a sealed record of each request ship in the same product.

Agents act on what they read, and no filter catches everything. So the decision belongs outside the model, and the record belongs to you.
The facts.
- Company
- Verosek
- Also appears as
- Verosek Solutions, on our booking link. That name has no separate website.
- Founded
- 2025, in Mumbai
- What it builds
- An AI + MCP Gateway and an AI red-teaming framework
- Field
- AI security: security for AI agents and the tools they call
- Open source
- The gateway is closed source. We maintain Assay, an open benchmark for MCP security, MIT-licensed. We presented it at Black Hat India Arsenal in October 2026.
- Website
- verosek.com is Verosek's only website.
- linkedin.com/company/verosek
- Talk to us
- Book a call
One gateway in four parts, and a red-teaming framework.
AI Gateway
AI Gateway is the model side of Verosek's AI + MCP Gateway: it routes each call your OpenAI, Anthropic or Gemini SDK already makes to one of your organisation's deployments, and chat calls fall back in priority order when one fails.
Shield
Shield is the scanning layer of Verosek's AI + MCP Gateway: once enabled, it runs each key's checks over prompts, replies, tool arguments and tool results.
MCP Tools
MCP Tools is the tool side of Verosek's AI + MCP Gateway: it checks every tool call against the agent key's rules before it runs.
Audit Vault
Audit Vault is the record side of Verosek's AI + MCP Gateway: it turns every request into a sealed trace of typed steps with secrets and personal data scrubbed.
Red Teaming
Red Teaming runs adversarial attacks against your AI agents and their MCP servers, with findings you can recompute yourself.

