lesavka/AGENTS.md

# Lesavka Agent Notes

## A/V Sync Probe And Lip-Sync Validation Checklist

Context: Google Meet testing on 2026-04-30 showed audio roughly 8 seconds behind video even though internal client/server telemetry reported fresh uplink packets. Treat this as a product correctness failure, not a calibration issue. Do not resume blind lip-sync tuning until the probe can explain where delay appears.

### Operating Principles
- Avoid hard-resetting USB, UVC, UAC, display managers, or remote hosts unless the user explicitly approves it.
- Prefer observation and reversible user-space probes before changing media pipelines.
- Treat Tethys-only SSH/device inspection as a development luxury, not a production dependency.
- Do not claim lip sync is fixed from internal telemetry alone; require end-to-end device-level evidence.
- Keep this checklist updated as work lands.

### Phase 1: Build The Probe
- [x] Create this tracked checklist in `AGENTS.md`.
- [x] Inventory existing `client/src/sync_probe/` code and decide what can be reused.
  - Reuse the existing synthetic beacon in `client/src/sync_probe/`.
  - Reuse the existing Tethys capture harness in `scripts/manual/run_upstream_av_sync.sh`.
  - Reuse and extend `lesavka-sync-analyze`; current gap is structured evidence output, not capture generation.
- [x] Define the phase-1 output contract:
  - [x] `report.json`
  - [x] `report.txt`
  - [x] per-event rows with event id, video time, audio time, skew, and confidence
  - [x] pass/fail verdict using preferred/acceptable/catastrophic thresholds
- [x] Add a deterministic local sync beacon source:
  - [x] video flash pattern with event identity or cadence
  - [x] simultaneous audio click/beep
  - [x] stable event schedule suitable for automated detection
- [x] Add a Tethys-side capture probe:
  - [x] capture Lesavka UVC video device
  - [x] capture Lesavka UAC microphone device
  - [x] record enough raw evidence for debugging when detection fails
  - [x] detect video flashes
  - [x] detect audio clicks
  - [x] pair events and compute skew
- [x] Add a runner that can launch or instruct the Tethys probe safely over SSH without rebooting or restarting the desktop.
- [x] Store probe artifacts under `/tmp/lesavka-sync-probe-*` by default.
- [x] Keep the probe usable without Google Meet first; Google Meet validation is a later application-level check.

### Phase 2: Use Probe To Root-Cause Desync
- [x] Run probe through direct Lesavka UVC/UAC devices on Tethys.
  - First live run reached the devices but exposed analyzer/tooling gaps instead of a valid skew report.
  - Fixed the manual probe tunnel to preserve HTTPS/mTLS through SSH (`LESAVKA_SERVER_SCHEME=https`, `LESAVKA_TLS_DOMAIN=lesavka-server`).
  - Fixed analyzer handling for MJPEG captures whose FFprobe metadata over-reports frames versus decodable video frames.
- [x] Compare client-generated event times against Tethys-observed times.
  - The preserved Tethys capture had 323 decodable frames with constant brightness, so no video flash reached UVC.
  - Server logs show the probe entered a stale upstream session and dropped audio as ~326 seconds late.
- [x] Identify whether delay appears before server planning, at server UAC sink, at UVC helper, inside Tethys device capture, or inside browser/WebRTC.
  - Current root cause is server planning/session lifecycle, before UVC/UAC sink output.
  - A previous one-sided microphone session started at 2026-04-30T22:59:52Z; the new probe at 2026-05-01T00:57:08Z inherited its stale playout epoch.
- [x] Add diagnostics for whichever stage is hiding delay.
  - Existing server lifecycle/planning logs were enough to isolate this run; next gate should preserve these as structured artifacts.
- [x] Do not tune calibration offsets until gross backlog is ruled out.
  - No calibration offsets were changed during the stale-session investigation.
  - Current evidence points at lifecycle/session planning, not an offset problem.

### Phase 3: Fix Lesavka With Evidence
- [x] If stale upstream lifecycle is confirmed, reset shared A/V timing anchors when a new stream replaces an existing owner.
  - Added a lifecycle guard so normal camera/microphone stream replacement clears stale shared timing anchors before re-pairing.
  - Kept soft microphone recovery intentionally separate so it supersedes the mic owner without disturbing an active healthy camera/shared clock.
  - Added regression coverage for stale timing-anchor replacement and soft microphone recovery preservation.
- [ ] If UAC sink backlog is confirmed, make UAC output freshness-bounded.
- [ ] If audio progress is marked too early, move/augment progress telemetry to reflect actual sink emission readiness.
- [ ] If UVC and UAC are using incompatible freshness semantics, unify them behind one live-media policy.
- [ ] If browser/WebRTC adds delay after devices are already synced, document the application boundary and add browser-specific mitigation or guidance.

### Phase 4: Gate And Release Criteria
- [x] Add deterministic unit/integration tests for probe analysis logic.
- [x] Add a hardware-in-the-loop/manual gate artifact schema for real Tethys probe runs.
- [x] Update `scripts/ci/media_reliability_gate.sh` to report probe evidence when present.
  - Gate now reads `LESAVKA_SYNC_PROBE_REPORT_JSON`, `LESAVKA_SYNC_PROBE_REPORT_DIR`, or `target/media-reliability-gate/sync-probe/report.json`.
  - Gate emits sync-probe verdict/check metrics, skew metrics, event counts, and a verdict info metric.
- [x] Require a fresh probe report before declaring lip sync fixed.
  - Gate now supports `LESAVKA_REQUIRE_SYNC_PROBE=1`, which fails media reliability when a valid passing probe report is absent.
  - Product/release judgment still requires a new live Theia/Tethys probe after the lifecycle fix is installed.
- [ ] Suggested thresholds:
  - [x] preferred: p95 skew <= 35 ms
  - [x] acceptable: p95 skew <= 80 ms
  - [x] gross failure: sustained skew > 250 ms
  - [x] catastrophic failure: any sustained skew near or above 1000 ms

### Open Questions
- [x] Decide whether the phase-1 beacon should run as a separate binary, a hidden client mode, or both.
- [x] Decide whether Tethys probe should be Rust-only, shell plus GStreamer, or a hybrid.
- [ ] Confirm whether sudo/Vault access is available for installing missing probe dependencies on Theia/Tethys.
  - Non-sudo server journal inspection worked; noninteractive sudo over SSH still needs an explicit TTY/password path.

### Validation Evidence
- [x] `cargo test -p lesavka_server upstream_media_runtime::tests::lifecycle`
- [x] `cargo test -p lesavka_client sync_probe::analyze`
- [x] `cargo test -p lesavka_testing upstream_sync_script_tunnels_auto_server_addr_through_ssh`
- [x] `bash -n scripts/ci/media_reliability_gate.sh`
- [x] `cargo test -p lesavka_testing media_reliability_gate_reports_direct_sync_probe_evidence`
- [x] `LESAVKA_REQUIRE_SYNC_PROBE=1 ./scripts/ci/media_reliability_gate.sh`
  - Used a synthetic passing report at `target/media-reliability-gate/sync-probe/report.json` to verify gate parsing/enforcement.
  - This validates CI glue only; a real Theia/Tethys probe is still required for product judgment.

## Real Upstream Lip-Sync Fix Checklist

Context: the mirrored browser probe finally reproduced the real failure class on 2026-05-01:
`activity_start_delta_ms=+9591.1`. This means the end-to-end browser-visible path can still start video far ahead of audio. The fix target is not silence in the logs; it is a freshness-first A/V uplink whose startup can heal briefly but cannot drift into seconds of skew.

### Acceptance Criteria
- [ ] Mirrored browser probe passes with `activity_start_delta_ms <= 1000`.
- [ ] Steady-state preferred sync: median skew within `35 ms`.
- [ ] Steady-state acceptable sync: p95 absolute skew within `80 ms`.
- [ ] Any sustained or startup A/V split near `1000 ms` remains a hard failure.
- [ ] No stale audio backlog is ever drained into UAC to catch up.
- [ ] No stale video backlog is ever drained into UVC to catch up.
- [ ] Google Meet manual testing agrees with the mirrored probe instead of revealing hidden seconds-scale skew.

### Phase 0: Keep The Probe Honest
- [x] Split raw activity-start fields from filtered/coded paired-pulse fields in probe reports.
- [x] Print explicit raw first-video and first-audio timestamps in `report.txt`.
- [x] Root-cause the 0.16.17 `raw_first_video_activity_s=0.000` artifact as the mirrored probe counting its own bright pre-start positioning card.
- [x] Make the mirrored stimulus pre-start screen dark/dim so only real flash pulses can be detected as video activity.
- [x] Add analyzer coverage proving dim pre-start positioning frames are ignored.
- [x] Replace generic light/dark mirrored flashes with color-coded event IDs.
- [x] Make mirrored audio pulses unique by the same event ID via pulse width plus tone frequency.
- [x] Teach the analyzer to decode mirrored video event IDs from color, not grayscale brightness.
- [x] Tighten real-camera color matching after 0.16.18 accepted washed-out brown/gray remnants as red/yellow events.
- [x] Preserve raw activity-start timing before cadence cleanup in coded reports.
- [x] Merge short audio envelope dropouts inside one coded pulse so a single tone burst cannot become two fake events.
- [x] Add diagnostic coded-pair correlation so stable large skew reports as measured failure instead of `not enough pairs`.
- [x] Make coded mirrored verdicts/calibration use matched coded pulses as authority; raw activity-start deltas are reported separately unless they agree with the coded pairs.
- [x] Print unpaired video/audio onsets in the human report so missed coded pulses are visible during probe triage.
- [ ] Keep the mirrored browser probe as the release/blocking upstream A/V gate.
- [ ] Keep the old raw-device probe as a lower-level diagnostic only.

### Phase 1: Stop One-Sided Startup Drift
- [x] Default upstream planning must require both camera and microphone before live playout.
- [x] One-sided playout may only happen through an explicit compatibility override.
- [x] While pairing is overdue, keep replacing the waiting-side anchor with fresh packets instead of preserving stale startup anchors.
- [x] While awaiting the peer stream, keep only fresh pending camera packets.
- [x] While awaiting the peer stream, keep only fresh pending microphone packets.
- [x] Send the latest camera packet from the client uplink queue instead of draining old-but-not-yet-stale video backlog.
- [x] Add tests proving the pairing window no longer expires into one-sided playout by default.
- [x] Add tests proving the explicit one-sided override still works for intentional single-stream scenarios.

### Phase 2: Bound UAC Freshness
- [x] Configure UAC `appsrc` as non-blocking and bounded.
- [x] Log and drop UAC appsrc push failures instead of treating enqueue as guaranteed playback.
- [x] Raise calibration offset limits enough to cover the measured MJPEG/UVC path delta without rejecting probe corrections.
- [x] Update the MJPEG/UVC factory audio baseline from the old `-45ms`/`+720ms` values to `+1260ms` as the mirrored probe exposes the fresh UAC-vs-UVC path delta.
- [x] Migrate untouched legacy `-45ms` factory/env calibration files on load so old installs actually receive the new baseline.
- [x] Make the video/audio-master wait offset-aware so a positive audio playout delay does not freeze UVC video while UAC sleeps before emission.
- [ ] Flush/stop UAC cleanly on session close, replacement, and recovery.
- [x] Add tests or contract coverage for bounded UAC settings where practical.

### Phase 3: Add Real Timing Evidence
- [ ] Add server timing counters for first camera packet, first mic packet, first UVC write, and first UAC push per session.
- [ ] Add dropped-stale audio/video counters to diagnostics.
- [ ] Add a concise health explanation when startup pairing exceeds the healing window.
- [ ] Surface `Starting`, `Healing`, `Flowing`, `Lagging`, `Dropping`, and `Stale` states in chips/diagnostics from real path evidence.

### Phase 4: Recovery And Mid-Session Changes
- [ ] Make device changes trigger soft-pause, stream replacement, queue flush, and re-pairing.
- [ ] Keep recovery soft-first; reserve hard UVC/UAC gadget rebuilds for explicit guarded recoveries.
- [ ] Add cooldown/state guards so recovery buttons cannot wedge Theia.
- [ ] Ensure disconnect closes all client/server media tasks for the session.

### Phase 5: Verification Loop
- [x] Run focused upstream runtime tests.
- [x] Run server/client media contract tests.
- [x] Run `cargo check` for touched packages.
- [x] Bump version for the fix release.
- [x] Run the mirrored browser probe on installed client/server.
  - 0.16.17 still failed: reported `activity_start_delta_ms=+6735.0`, but `raw_first_video_activity_s=0.000` exposed a probe false-positive from the pre-start screen. Paired pulses still showed real steady-state skew (`p95=411.8 ms`, `median=-99.0 ms`), so the product remains unfixed.
  - 0.16.18 captured real colored/audio-coded events but the analyzer still bailed with `need at least 3 matching coded pulse pairs; saw 1`. Replaying that artifact after analyzer hardening now reports `gross_failure`: 16/16 coded pairs, p95 `775.7 ms`, activity start `-766.4 ms`, and drift `-2.8 ms`; the failure is stable audio-ahead/video-late skew, not random detector noise.
  - 0.16.19 changes the shipped MJPEG/UVC audio playout baseline to `+720ms`; the next mirrored browser probe should move the measured median from about `-766ms` toward roughly `-46ms` before fine calibration.
  - 0.16.19 mirrored browser probe did not move the measured skew: p95 `885.7 ms`, median `-788.4 ms`, activity start `-659.1 ms`, drift `-81.2 ms`. SSH inspection showed Theia was on commit `c348597`, but `/etc/lesavka/server.env` still contained `LESAVKA_UPSTREAM_AUDIO_PLAYOUT_OFFSET_US=-45000`; the new `+720ms` baseline was not actually installed. Patch the installer to migrate leaked legacy ambient `-45000` to `+720000` unless `LESAVKA_INSTALL_UPSTREAM_AUDIO_PLAYOUT_OFFSET_US` explicitly asks for the legacy value.
  - 0.16.20 installed the `+720ms` offset (`/etc/lesavka/server.env` had `LESAVKA_UPSTREAM_AUDIO_PLAYOUT_OFFSET_US=720000`), but the mirrored browser capture contained no recognizable color pulses. Theia server logs showed repeated `upstream video frame dropped because the audio master never caught up inside the pairing window`; UVC was effectively starved by the positive audio delay instead of flowing delayed-but-fresh frames.
  - 0.16.21 makes that wait offset-aware and adds a regression test proving a configured positive audio delay does not freeze UVC video while UAC sleeps before playout.
  - Replaying the 0.16.21 artifact after 0.16.22 analyzer hardening changes the verdict from false `catastrophic_failure` to `gross_failure`: p95 `273.8 ms`, median `-188.4 ms`, 7 paired coded pulses. The raw activity-start delta (`-3620.7 ms`) is still printed, but it is ignored for verdict/calibration because it disagrees with coded pairs by `3432.3 ms`; unpaired video/audio onsets are printed for triage.
  - 0.16.22 live mirrored run still failed with p95 `433.7 ms`, median `-359.4 ms`, and 5 paired coded pulses. Client telemetry showed camera uplink `latest_age_ms` repeatedly around `300-350 ms`, matching the measured skew; patch 0.16.23 to make video queues latest-only instead of draining stale-but-under-budget backlog.
  - 0.16.23 local validation passed for fresh-queue behavior, uplink/probe freshness contracts, sync analyzer tests, client/server binary checks, and whitespace checks.
  - 0.16.23 live mirrored run improved to p95 `215.2 ms`, median `+142.2 ms`, 13 paired coded pulses, and raw activity alignment within `6.6 ms` of coded pairs. Patch 0.16.24 makes the probe print local client and remote server versions before capture so every run records what was actually tested.
  - 0.16.24 live mirrored run improved again to p95 `168.4 ms`, median `-19.1 ms`, 11 paired coded pulses, but still failed because individual paired pulses bounced between about `-168 ms` and `+45 ms`. Client logs showed the microphone uplink queue still accumulating depth `16`; patch 0.16.25 makes microphone uplink queues latest-only too so stale audio PTS cannot continue acting as the server timing master under backpressure.
  - 0.16.25 removed the client mic backlog but exposed a stable hardware/browser path delta: p95 `557.3 ms`, median `-540.5 ms`, drift `+9.0 ms`, and fresh mic delivery ages around `2-10 ms`. Patch 0.16.26 raises the MJPEG/UVC factory audio delay to `+1260 ms` and expands the calibration clamp so this stable offset can actually be corrected instead of rejected.
- [ ] Re-run the mirrored browser probe after the pre-start false-positive fix.
- [ ] Run Google Meet manual validation.

## 0.17.0 Tyrannical Upstream Playout Checklist

Context: 0.16.x proved that queue tweaks and static calibration cannot guarantee lip sync. 0.17.0 changes the upstream contract: the server planner is authoritative, audio is the master, video follows by timestamp or freezes, and freshness wins over smooth-but-wrong playback.

### Hard Product Invariants
- [x] No normal live upstream playout may be more than 1000 ms behind the freshest known client capture frontier.
- [x] Video may not advance outside the audio playhead sync budget.
- [x] Audio should be continuous when possible, but stale audio must be dropped/skipped rather than drained.
- [x] Missing or late video should freeze/stutter instead of pulling audio backward.
- [x] Startup may be ugly, but must either enter fresh synced live mode or declare failure within 60 seconds.
- [x] Healing may be visible, but must prevent persistent seconds-scale skew.
- [x] Calibration may fine-tune sub-frame offsets only; it must not be required to rescue seconds-scale desync.

- [x] Bump Lesavka to 0.17.0 because this is a media-contract change, not a patch tune.
- [x] Bump patch follow-ups to 0.17.1 instead of reporting `version+revision` as the release version.
- [x] Add planner policy config: max live lag, max skew, startup timeout, target playout delay, and healing cooldown.
- [x] Reset 0.17 defaults so shipped audio/video offsets do not intentionally exceed the freshness budget.
- [x] Track latest camera/audio input timestamps in the server planner.
- [x] Track actual planned/emitted audio and video playheads.
- [x] Enforce audio as master: stale audio is dropped/skipped; it does not drain backlog.
- [x] Enforce video follower behavior: frames ahead of audio wait; frames too far behind audio are dropped so the previous frame freezes.
- [x] Re-anchor continuously when the live playhead falls outside the freshness budget, not only once at startup.
- [x] Keep startup paired-only by default and fail visibly after the startup timeout.
- [x] Add planner phases and counters to diagnostics/logs: acquiring, syncing, live, healing, failed; stale drops, skew drops, freshness heals, video freezes.
- [x] Keep UVC/UAC sinks as dumb consumers of planner-approved packets.
- [x] Update tests to prove stale media cannot be emitted and video cannot outrun/lag audio beyond policy.
- [x] Update manual/probe diagnostics so 0.17 reports the planner state being tested.

### Validation Targets Before Human Test
- [x] Unit tests for startup pairing, stale audio drop, stale video drop, reanchor, startup timeout, audio-master/video-follower rules.
- [x] Contract tests for installer defaults and version reporting.
- [x] `cargo check -p lesavka_client -p lesavka_server --bins`.
- [x] Focused `lesavka_testing` media/runtime contracts.
- [x] Only after all of the above, run the mirrored browser probe.

### Progress Log
- 2026-05-01: Added 0.17 planner defaults (`350ms` target playout, `1000ms` max live lag, `60000ms` startup timeout, `80000us` pair slack), reset MJPEG audio factory offset to `0`, and migrated old `-45ms`, `+720ms`, and `+1260ms` untouched baselines.
- 2026-05-01: Server planner now tracks latest input frontier, presented audio/video playheads, phase, stale drops, skew drops, reanchors, startup timeouts, and freezes.
- 2026-05-01: Runtime tests green for stale audio drop, stale video drop, audio-master/video-follower freeze, repeated reanchor, paired startup timeout, and planner snapshot basics: `cargo test -p lesavka_server upstream_media_runtime::tests -- --nocapture`.
- 2026-05-01: Added `GetUpstreamSync` RPC, `lesavka-relayctl upstream-sync`, launcher diagnostics text, and mirrored-probe before/after planner snapshots so 0.17 probe runs report the exact planner state under test.
- 2026-05-01: Validation green: `cargo test -p lesavka_server --lib --bins`, `cargo test -p lesavka_testing`, `cargo test -p lesavka_client --bins --lib`, and targeted installer/RPC/layout contracts.
- 2026-05-01: First installed 0.17.0 mirrored browser probe on client/server commit `3920e0a` failed honestly: planner reported fresh live state (`live_lag_ms=10`, `skew_ms=+20.7`) but browser-observed paired pulses showed audio late by median `+349.1ms`, p95 `429.1ms`, with 6 video freezes/skew drops. Replayed artifact after analyzer hardening now reports `gross_failure` instead of false raw-start `catastrophic_failure`.
- 2026-05-01: Patch follow-up models the observed MJPEG/UVC browser egress delta by defaulting video playout offset to `+350ms` and preserving the 1s freshness ceiling. Raw activity-start evidence is now ignored for verdict/calibration when it disagrees with paired pulses that are already failing directly. Existing early-0.17 `audio=0/video=0` factory/env calibration files migrate to the new `video=+350ms` default on load.
- 2026-05-01: Release identity cleanup: bumped the patched build to clean semver `0.17.1`; probe attribution now prints `client_version`/`server_version` separately from `client_revision`/`server_revision` and refuses old `client_full_version` output.
- 2026-05-01: 0.17.1 mirrored probe failed with video about `1.18-1.31s` behind audio and 761 planner video freezes. Root cause candidate: the client rebaser forced independent camera/mic pipelines onto one first-packet capture base, so a later-starting camera path was timestamped too early and looked permanently behind audio. Patch 0.17.2 anchors each stream to the shared monotonic clock at its own first packet time.
- 2026-05-02: 0.17.2 mirrored probe and Google Meet test showed major improvement but persistent sub-second late video. Root cause follow-up: the temporary `+350ms` factory MJPEG video playout offset matched the observed browser skew and also made the server skew guard freeze video against its own offset. Patch 0.17.3 restores factory video offset to `0ms`, migrates untouched `+350ms` install/calibration defaults back to `0ms`, and makes the skew guard offset-aware for intentional site calibration.
- 2026-05-02: 0.17.3 Google Meet manual test improved to roughly sub-second/near-quarter-second lip sync, but the mirrored analyzer could not pair pulses and the user still heard choppy background audio. Client logs showed Pulse microphone packets arriving unevenly with ages around `90-240ms`; patch 0.17.4 lowers Pulse mic `buffer-time`/`latency-time`, bounds the mic queue/appsink, and keeps mirrored-probe after-run planner diagnostics even when analysis fails.