FyVault
Emergency Access

It's 3am. Production is down. Now what?

Your payment API is returning 500s. Customers are churning by the minute. The engineer who set up the Stripe credentials is on vacation. The rotation policy changed the password last week. You are locked out of your own infrastructure.

Break-Glass Flow

Break the glass. Fix the problem. Access revokes itself.

Outage hits production

Engineer runs break-glass

Incident reason logged

Scoped token minted

Security team alerted

Access self-destructs

When seconds matter, paperwork doesn't

One command. The CLI mints a scoped token, alerts your security team, and starts the self-destruct timer. No approval chains. No Jira tickets. Just access.

Break-Glass Session
$ fyvault break-glass \
--env=production \
--reason="incident-2024-042: Stripe API returning 500s" \
--ttl=60
⚠ BREAK-GLASS SESSION ACTIVATED
Session: fvbg_m9k2a...
Environment: production
Reason: incident-2024-042: Stripe API returning 500s
Expires: 60 minutes
✓ #incidents-security notified via Slack
✓ PagerDuty webhook dispatched
Token: fvbg_m9k2a_sk_live_emergency...
⏱ Auto-revocation in 59:59
Safety Guardrails

Emergency access that doesn't compromise security

Self-destructing access

Every token has a hard ceiling of 4 hours. The server enforces the countdown. No renewals. No workarounds. Time's up, access is gone.

Mandatory incident context

No blank checks. Every break-glass session demands a typed reason and an incident ID before a single byte of credential data is exposed.

Real-time alerting

Slack, webhooks, and email fire the instant the glass breaks. Your security team knows who, when, and why before the engineer even reads the secret.

Immutable audit log

Who triggered it. What they accessed. When access expired. Why they needed it. Permanently recorded. Tamper-proof. Ready for your next compliance audit.

Every team has a 3am story. This is how it ends differently.

The 3am nightmare
  • Call the one person who knows the password — they're asleep in another timezone
  • Post credentials in a Slack DM and hope nobody screenshots it
  • No record of who accessed what during the incident
  • Forget to revoke emergency access three weeks later
With break-glass
  • One command gets you scoped, read-only access in seconds
  • Token auto-revokes when the clock runs out — no cleanup needed
  • Every action logged from the moment the glass breaks to the moment access dies
  • Slack and webhook alerts so security knows before you've even read the secret
SDK

Wire it into your incident playbook

PagerDuty fires. Your runbook calls the SDK. A scoped token is minted before the on-call engineer even opens their laptop.

breakGlass.ts
const session = await fv.breakGlass.create({
reason: "incident-2024-042: Stripe 500s",
environment: "production",
ttlMinutes: 60,
});
// Token self-destructs in 60 minutes
// Security team alerted before you read a single secret

Emergency access in four steps

From incident to auto-revoke. No manual cleanup. No forgotten tokens.

Incident
Break-glass command
Time-limited token
Auto-revoke ⏱
Python SDK

Automate break-glass from Python

Trigger emergency access sessions programmatically from your incident runbooks.

break_glass.py
session = fv.break_glass.create(
    reason="incident-2024-042",
    environment="production",
    ttl_minutes=60,
)

# Token self-destructs in 60 minutes
# Security team alerted instantly

When the building is on fire, you break the glass. You don't fill out a form.

Give your team a way to act in seconds, not hours. One command. Time-limited. Fully audited. The access revokes itself so you never have to remember.