* fix: allow Linux interface discovery in embedded media service
* feat: expand local diagnostic reports
* research: validate independent Browser encoding producers
* feat: prototype Browser pool ownership and record remaining media gates
* feat: refine Browser shared encoders and native output reuse
* Refine carrier costs and simplify release entry points
* Close media phase and prepare piik design handoff
Embed STUN and SFU in the shared Go process, reuse native encoding pipelines with independent adaptation, and complete Client lifecycle, settings, diagnostics and packaging fixes. Retain standard Browser senders after measured encoded-reuse limitations.
Complete the Native Client source/audio boundary, strict v5 loopback contract, Browser media lifecycle handling, and non-blocking release checks. Keep unverified platform/physical gates explicit; do not rewrite the Go Server.
Load only the active client route, expose resolved live codec state, and add exact depth-aware browser quality evidence with current-UI quality-control gates.
Modularize durable truth, bound always-loaded instructions, and keep domain-document size checks advisory. Local cross-platform repository hygiene checks pass; GitHub-hosted steps were not started because of the account billing state.
Quarantine disputed TODO and routing claims, record current production facts, and define the canonical workspace restoration order before any further implementation.
Preserve exact selected edges across unrelated route revisions, demote corroborated bad relay parents, and parameterize the bounded endpoint cap while retaining the production default of two.
Keep local media validation focused on topology and recovery correctness, move Native packaging to a manual release boundary, and persist queued-feedback workflow guidance.
AAA