| name | distributed-voice-plane-governance |
| description | Govern the local-first Penny speech plane across ASR, TTS, endpoint voice behavior, and backend action routing. Use when Shockwave, AvatarCore, VRM Sandbox, or control-plane voice contracts shift together. |
Distributed Voice Plane Governance
Use this skill when the Penny voice fabric changes across multiple local repos.
Workflow
- Confirm the canonical speech-plane owners for ASR, TTS, and endpoint control.
- Inspect local loopback contracts before touching higher-level orchestration.
- Keep voice capture, transcript routing, backend action governance, and playback ownership aligned.
- Validate the speech loop with a bounded readiness or smoke sequence.
- Record any degraded or fallback voice paths explicitly.
Required Evidence
- ASR and TTS owner surfaces involved
- endpoint or backend routing rule touched
- bounded speech-loop validation note
- any fallback or degraded path still present
Guardrails
- Do not let endpoints bypass backend policy.
- Keep the speech plane local-first.
- Do not introduce direct cloud dependence as the canonical voice path.
- Fail closed if voice routing drifts from the documented owner ports or services.
Best-Fit Companion Skills
$voice-call$local-compute-usage$heterogeneous-stack-validation$skill-enforcer
Scope Boundary
Use this skill only for cross-repo speech-plane contracts and governance.
For one repo's TTS or ASR implementation details, route to the repo-specific voice or runtime skill.
