CISSP | A second set of eyes on your AI agent's permissions
I'm a CISSP with 10+ years in IT operations, incident response, and identity and access management. Every week I write RAIID, the Rogue AI Incident Digest, which tracks real cases of AI agents leaking data, escalating their own permissions, or acting without approval. I also built Cooren AF, a human-approval gate that sits in front of risky agent actions.
Book me if you've connected an AI agent or MCP server to real tools (email, databases, payments, deployments) and want a second opinion from security before something goes wrong. I review configurations and designs. I'll never ask for production credentials.
Walk me through your agent and its tools. You leave with your top 3 risks and the fix for each.
3 sessions + 2 asks spread over a month while you ship an agent: design review, pre-launch check, and post-launch review.
We pick the one action your agent must never take alone, like refunds, deletes, sends, or deploys, and design the approval step and audit log for it.
Walk me through your agent and its tools. You leave with your top 3 risks and the fix for each.
3 sessions + 2 asks spread over a month while you ship an agent: design review, pre-launch check, and post-launch review.
We pick the one action your agent must never take alone, like refunds, deletes, sends, or deploys, and design the approval step and audit log for it.