What you receive
- A bounded watch: An explicit list of your domains and host checks. No probing of third-party sites, exploits, or surprise changes.
- Evidence you can inspect: Deterministic findings, local SQLite history, a hash-chained audit log, and a daily HTML/JSON report. AI does not decide whether a finding is true.
- Your approval stays in the loop: CyberSENZ recommends fixes. It does not patch, block traffic, rotate credentials, or delete files on its own.
- A narrow agent connection: With an approved model route, your agent can read the guard’s status, findings, and report through a read-only MCP interface. Cloud routes without approval receive no security data. Monitoring continues if the agent or model is unavailable.
Current module coverage: Homepage availability and content changes, TLS certificate validity, and configured host PHP runtime versions. The wider host, vulnerability-feed, file-integrity, and connected-app reviews are planned expansions and are not included in the current module. Coverage is shown during onboarding, so you know what is and is not watched.
How installation works
- Choose your route. Add CyberSENZ to a DeepSenz Personal Agent installation, or request a standalone install for an eligible Hermes-based Personal Agent or PPC.
- Agree the scope. We record the domains, host, and checks you own and authorize. We confirm where each service runs; a managed-hosting dashboard is not the same as shell access to the host.
- Install and test. We place the module on your host, connect the read-only MCP tools to your agent where supported, confirm the scheduled checks and report, and send test alerts through both Telegram and email.
- Activate monitoring. Your service period starts after those checks pass. You receive the coverage list and instructions to review, acknowledge, or stop alerts.
Works with your agent
DeepSenz Personal Agent and Hermes Agent: CyberSENZ’s local MCP server and companion skill connect to Hermes. Detailed reads require an approved private model route; free cloud routes receive no security data. The monitoring process remains separate from Hermes.
Claude Code and Codex: Both support local MCP servers. The same read-only interface can be configured on a machine where CyberSENZ is installed; customer installations still require a connection test.
Grok Bot: Grok Bot documents command-based MCP connectors, so a local integration is possible on a supported computer. This has not been tested with CyberSENZ. Grok’s cloud connector would require a separately hosted, authenticated endpoint.
Meta Muse and Claude Duet: Direct CyberSENZ integration is not verified. Their cloud-agent environments do not have an installed CyberSENZ connector today. Ask us to assess a future connection rather than assuming compatibility.
Privacy and model costs
The current checks and report run locally on the monitoring host and make no paid model calls. Telegram and email alerts send only the asset name, severity, finding title, and ID to your chosen destinations. If your agent uses a free cloud model, CyberSENZ withholds security data by default. A future AI-written explanation may use a local model or an approved paid cloud model only after the route and cost are disclosed and private findings are redacted. Your agent’s memory is not copied into the guard.
Buying CyberSENZ Lite
With a DeepSenz Signature installation: Onboarding and the first 60 days of CyberSENZ Lite are included. After that, choose $9.99 per month or $99 for 12 months of scheduled defensive monitoring within your agreed scope.
Standalone add-on: $9.99 per month or $99 for 12 months of scheduled defensive monitoring for an eligible Personal Agent or PPC, after installation and activation. Hosting, a private network, and any optional paid model use are quoted separately before activation.
Checks run on a schedule while the monitoring host is online. CyberSENZ does not continuously inspect activity between checks.
Lite currently checks approved homepage availability and changes, TLS certificates, and any explicitly configured local PHP runtime. It saves a daily report on the monitoring host and sends urgent Telegram and email alerts. It does not yet send a routine daily all-clear message or protect web browsing. A broader managed service is planned, not offered today.
The service period starts only when the agreed checks, scheduled operation, report, Telegram and email alerts, and an independent health check pass the activation test. The purchase covers the written scope of monitoring and response guidance; it is not a guarantee that every attack will be detected or prevented.
Start with a short setup request. We confirm the scope and install CyberSENZ Lite, then send the Stripe payment link after monitoring is activated.
Frequently asked
Can I add it to an existing Hermes Personal Agent? Yes, if the host can run Python 3.11 or later and the agent can connect to its local MCP server. We verify access, permissions, alerts, and the exact checks before activation.
Can it fix a problem automatically? No. The current module reports and recommends. You decide what to change.
Is it entirely local? Detection, storage, and the standard report are local. Telegram and email delivery leave the host with a short, redacted alert. No model is required.
Does it replace antivirus, a firewall, or an incident-response professional? No. It adds a visible watch around the approved setup.