Launching today

aiSecurityWarden
Human approval gate for your AI agent's secrets and actions
1 follower
Human approval gate for your AI agent's secrets and actions
1 follower
aiSecurityWarden is a zero-knowledge secrets manager with a human approval gate for AI agents. Your agent asks before it uses a secret — you approve, deny, or deny with feedback from your phone. Nothing runs, no credential released, until a human says yes.






Free Options
Launch Team

AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted
Hey Product Hunt! 👋
We built aiSecurityWarden because we kept watching AI agents get handed raw API keys and OAuth tokens — one prompt-injection email away from someone draining your account.
aiSecurityWarden sits between your agent and your secrets:
The agent requests an action; YOU get an approval card on Telegram, Discord or ntfy
Approve, deny, or deny with a comment — the agent rewrites and resubmits
The secret is never in the agent's memory or logs (zero-knowledge; we can't read it either)
Same guardrail for your GitHub: the GitHub connector lets your agent read repos freely, but merges, pushes and issue/PR writes stop for your approval — a prompt-injected agent can't silently change your codebase.
The free plan includes every security feature — the paid tiers are just more projects and seats.
Would genuinely love feedback, especially from anyone running agents with production credentials. What's the scariest thing you've let an agent do autonomously?