Janee
Secrets management for AI agents
(Afrikaans: "yeeees... no" — the sound of a gatekeeper with opinions)
Your AI agents are running wild with your API keys.
We fix that.
The Problem
- AI agents need API access to be useful
- Current model: give them all the keys and hope they behave
- No audit trail of why something was accessed
- No kill switch when things go wrong
- One prompt injection away from disaster
The Solution
- Store your API keys in Janee — Stripe, Twilio, exchanges, databases
- Agent requests access with intent — "I need Stripe to check balance because user asked"
- LLM evaluates the request — Legit? Within policy? Suspicious?
- We proxy the call — Agent never sees the real key
- Everything logged with reasoning — Full audit trail
# Before (terrifying)
agent_config = {
"stripe_key": "sk_live_...", # 😱
"aws_secret": "...", # 💀
}
# After (sane)
janee_token = janee.request_access(
service="stripe",
reason="checking balance for revenue query"
)
# Agent gets scoped, audited access ✅
🚧 Status: Pre-MVP. We're building in public. First version ships in the next few weeks.