The pause is the security model.
Etenos is designed so that AI work is observable and reversible: local by default, gated before every side effect, sanitized on sync, and recorded in an audit trail you can review.
Six principles, enforced by design.
Local-first execution
The agent runtime runs on the developer's machine. Repository contents and command output are processed locally by default.
/01Explicit approvals before side effects
Six gates — branch, patch, validation shell, commit, push, PR — each hold and ask before anything mutates your repo or filesystem.
/02Sanitized control-plane sync
The cloud control plane receives sanitized activity summaries, not raw source or raw command output, unless an explicit future policy changes that.
/03GitHub token handling
Connects through a GitHub App installation or personal authorization. Tokens are scoped to the installation; provider API keys are never collected.
/04Agent registration & heartbeat
Each desktop agent registers with the control plane and emits a heartbeat, so an organization can see which agents are live and monitored.
/05Audit trail
Approvals, executions, policy changes, workspace-trust changes, and remote actions are recorded for review and export.
/06Nothing mutates without approval.
Six gates stand between the agent and your repository. Each one holds, shows exactly what will happen, and waits. Read-only stages — polling, scanning, planning, context — never touch your code.
GitHub token handling
Registration & monitoring
What we hold ourselves to in production.
Etenos services are built to production expectations — not development shortcuts.
On the roadmap — not claimed as live.
We're explicit about what isn't here yet. The following are enterprise roadmap items, not current guarantees. If your evaluation depends on any of them, talk to us about timelines.
Evaluate the model, not the marketing.
Read how the platform draws its data boundary, or bring your security team into a conversation.