Channels And Voice
X-Hub treats remote channels and voice as governed operator surfaces, not shadow control planes. External events should enter the Hub first, then be projected into trusted paired surfaces such as X-Terminal voice, mobile, or runner-style execution surfaces.
The Rule
All external-world events enter the Hub first.
The Hub owns:
- ingress authorization
- replay protection
- grant handling
- audit
- memory truth
- routing
- Supervisor-facing state projection
Remote Channels
The active channel direction centers on:
- Slack
- Telegram
- Feishu
- WhatsApp Cloud as an explicit later extension
The important part is not just channel coverage. It is that these channels stay inside Hub governance instead of becoming direct command runtimes.
Safe Onboarding
The onboarding direction for new remote ingress is:
- unknown ingress enters discovery or quarantine
- local admin approves once on a trusted management surface
- Hub writes identity and channel binding
- Hub runs a low-risk first smoke
- only then does the path become a normal governed ingress
That is a much safer default than allowing every new chat surface to become a live trusted control path immediately.
Voice As A Paired Surface
X-Terminal voice is not just dictation. It is the paired high-trust interaction surface that turns Hub-governed state into something the operator can hear, confirm, and continue.
The active preview direction already includes:
- proactive Supervisor briefing
- Hub-issued voice challenge state
- repeat and cancel semantics
- mobile-confirmation latch handling
- source-aware pending-grant targeting
- post-action re-briefing back through the Hub path
Why Source Awareness Matters
If multiple grants or remote requests are pending, voice should not guess.
That is why the voice flow is moving toward:
- source-aware pending-grant summaries
- remote-channel-aware targeting
- fail-closed behavior when grant selection is ambiguous
This is one of the places where usability and governance have to reinforce each other.
The Product Shape
The desired operator experience looks like this:
- remote channel triggers a governed request
- Hub classifies and holds the decision path
- X-Terminal voice briefs the operator
- the operator approves or rejects through a guided challenge flow
- the system re-briefs using the same Hub truth path
That makes voice and channels part of one governed interaction chain instead of separate ad hoc surfaces.