Updated
Where Nionx Cloud is going.
What works today, what I'm building now, and everything between here and version 1.0. No dates on purpose: I'd rather ship than promise.
Shipped
Working today in the closed beta.
Encrypted secrets, Shipped
AES-256-GCM envelope encryption. Values are write-only, with version history and one-click rollback.
Workspaces, projects and environments, Shipped
Development, staging, production and your own environments, with owner, admin, developer and viewer roles.
API tokens, Shipped
Read-only, scoped to one environment and always expiring. Every read is in the audit log.
Two-factor sign-in and confirmation, Shipped
Authenticator app codes, and your password and code again before deleting, revoking or changing access.
Invitations and audit log, Shipped
Invite teammates by email and see every change and every read, without the values.
AI agent connection (MCP), Shipped
Claude Code and other agents define secrets and check what is set. You type values in on a one-time link; the agent never sees them.
Public status page, Shipped
status.nionx.cloud: uptime, incidents and response times, checked every 2 minutes, plus a live measurement from your own browser.
Dashboard for beta testers, Shipped
Live at dash.nionx.cloud for invited testers, with verification and invitation emails.
Building now
What's needed to open the beta to testers.
Managed key storage, In progress
Encryption keys move to a managed key service (AWS KMS), so stored values are protected outside local testing.
Command-line tool, In progress
One command connects a project. Another starts your app with its secrets loaded, without writing a .env file.
npx nionx init
Next
Planned after the first testers.
Secrets cached in your app, Planned
The CLI and SDK keep secrets in memory and refresh them in the background, so reads after the first take no time at all.
Leak protection, Planned
Nionx tokens found in public code are revoked automatically.
Documentation, Planned
Guides for the API, the command-line tool and AI agents.
Trust
Don't take my word for it.
A secrets tool asks for trust. These are the ways you'll be able to check instead.
- With the first CLI release
Open-source command-line tool, planned
Read exactly what runs on your machine.
- Before the open beta
Public security design, planned
The threat model and encryption design, published for anyone to check.
- Before the open beta
Independent security review, planned
An outside review before the open beta, with a public summary of what was found and fixed.
Open beta
Anyone can sign up, without an invite.
Fast reads worldwide, Planned
Database copies in the US and Asia for quick reads everywhere, with revoked tokens still checked against the main database, so revoking stays instant.
Plus and Max plans, Planned
Plus at $5 a month for much higher limits, Max at $10 for heavy use with the cheapest extra API calls. The Free plan stays.
Open sign-up, Planned
No invite needed. People on the waitlist get in first, then sign-up opens to everyone.
Integrations, Planned
Load secrets in GitHub Actions and on popular hosting platforms without copying them by hand.
No stored tokens for supported platforms, Planned
Apps on supported clouds prove who they are and read their secrets without a long-lived token to leak.
More AI tools, Planned
The agent connection works with more tools than Claude Code, like Cursor and Codex.
Your data, your call, Planned
Export everything, or delete your account and all its data yourself.
Tested backups, Planned
Encrypted backups, with a restore procedure that's actually practised.
Public release 1.0
Ready for real production secrets.
Production ready, Planned
The "testing only" label goes away once everything above is done and the security review is clean.
Secret and key rotation, Planned
Rotate a secret everywhere it's used in one step, and rotate encryption keys on a schedule.
Enterprise plan, Planned
For companies: single sign-on, custom key management and audit log export, with custom pricing.