Security model

How Opsaido protects your systems, logins and data — layer by layer, each assuming the others can fail.

Opsaido can change real systems, so it’s built on defence in depth: every layer assumes the others can fail.

1 · PeopleSign-in, two-step verification, single sign-on, roles
2 · ApprovalsReads run; changes wait for a human; dangerous actions need a separate permission
3 · The assistantEvery step checked against a schema and an allow-list before it runs
4 · SecretsEncrypted at rest, never sent to the browser or the AI
5 · ConnectionsPinned host keys, TLS, dedicated least-privilege accounts
6 · PlatformHardened container, private database, strict browser headers
7 · AuditEvery action, approval and session recorded

What the AI can and can’t do

The AI canThe AI can’t
Choose which tool to use nextRun anything you haven’t connected
Read results to explain themSee passwords, keys or tokens
Propose changesMake a change without your approval
Plan and rehearse a dry runGo off-plan after you approve it — any variance asks again
Use tools an admin has switched off, even if it asks for them

The AI’s output is treated as untrusted. Every tool call is checked against its schema, and names such as hosts, users and services are checked against strict patterns, so a reply from the model can’t smuggle in shell, PowerShell or LDAP commands.

Prompt injection

Content from outside — emails, web pages, log lines — could contain text written to trick an AI. Opsaido limits the damage:

ProtectionEffect
Outside content is marked as dataEmails and similar content reach the AI labelled as untrusted, with the instruction never to act on what they say
Approvals still applyEven if the AI were fooled, a change still needs a person to approve it
Dangerous actions are separateDeleting, powering off or restoring needs a permission most people don’t have
Everything is auditedYou can see exactly what was asked, proposed, approved and run

Logins and secrets

WhatHow it’s protected
Connector logins, API keys, bot tokensEncrypted at rest with a key derived from your SECRET_KEY; decrypted only for the moment a connection is made
Sent to the browser?Never. Forms show “stored — leave blank to keep”
Sent to the AI?Never. The AI sees results, not how Opsaido signed in
User passwordsArgon2id hashes, 12+ characters, rate-limited sign-in
Sessions256-bit random tokens stored only as hashes; HttpOnly, SameSite, Secure cookies
Two-step codesTOTP with single-use recovery codes and replay protection
BackupsEncrypted with a passphrase you choose

Keep SECRET_KEY safe and backed up. Without it, stored logins can’t be decrypted, and anyone who has it plus a copy of the database could read them.

Connections

ProtectionEffect
SSH host keys pinnedThe key is saved on the first test. A changed key is refused, which stops a machine pretending to be your server
TLS verifiedCertificates are checked. Add your own CA in certs/ca/ rather than turning checks off
Least privilegeEach connector uses its own account with only the rights you give it; read-only is enough to monitor
GatewaysA jump host needs no admin rights and holds no logins for the targets

The platform

LayerHardening
ContainerRuns as non-root with a read-only filesystem, all Linux capabilities dropped and no-new-privileges
DatabaseOn an internal network only; not reachable from outside
Remote desktop serviceReachable only by Opsaido itself
BrowserStrict Content-Security-Policy, X-Frame-Options: DENY, nosniff, HSTS over HTTPS
RequestsChanges need a custom header (CSRF protection); live connections check their origin

What leaves your network

Goes toWhatWhen
Your AI providerYour question and the results of the stepsEach request. Use a local model to keep this on your network
The licence serverLicence ID, installation ID, version and counts only (number of machines)A periodic check-in. Offline licences never check in
Services you set upAlerts and replies to your notification channels (email, Telegram, Slack…); requests to cloud services you connect (Microsoft 365, Cloudflare…)Only when you add them
Nothing elseNo telemetry, no analytics, no host names, no logins—

For the strictest setups, run a local model (Ollama, vLLM, LM Studio) with an offline licence, and only on-network notification channels. Then nothing leaves your network at all.

Your part

DoWhy
Turn on two-step verification for every adminA stolen password alone isn’t enough
Give connectors least-privilege accountsLimits what any mistake can touch
Keep dangerous-action approval for a few peopleDeleting and powering off stay rare and deliberate
Put Opsaido behind HTTPS with your own certificateProtects sign-in and sessions on the network
Back up SECRET_KEY and the databaseSo you can restore — and so you know where copies are
Review the audit log now and thenSpot anything unexpected early