* Rename face model to face recognizer
* Refactor face detection into own module
* Return face and face landmarks
* Align faces with 5 points instead of just the eyes
* Add landmark validation to throw out images which do not fit a landmark
* Fix circular import and lock face detector for concurrent runs across threads
* Fix mypy
* Support multi resolution exports
* Fix decoder text
* Add dropdown and ability to select export stream selection
* Fix for review comments
* Fix mypy
* Cleanup wording
* rename migration
* use ASC composite index for event camera and start_time
the planner still picks for newest-first single-camera queries but doesn't slow down full per-camera scans on a cold cache
* pause playback while the timeline range handles are up
* reduce multi camera seeded export range to 30m
allows both handlebars to fit within most desktop windows
* Add additional indicies on event and review tables. Every events or timeline endpoint filters on event start time and camera, this should speed things up by avoiding a range scan on the table.
* Rewrite to use a CTE to leverage speedups by using sqllite internal optimization to do a single query instead of a starter query to get distinct labels and a subsequent loop of querys per distinct event labels.
Frigate is currently shipping sqlite 3.46.1, which is above the minimum version 3.25 needed for CTEs.
* Collapse a few sequential queries into a single one.
* Use peewee instead of rw sql for the CTE query.
* Slightly simplify review logic and avoid duplicating the json response for empty review IDs.
* Rerun ruff formatting.
* Remove 2x unnecessary index on reviewsegment, remove reference to prior code implementation in comment in event.py
* Editor fail, re-ruff format.
* Remove the CTE and restore the generator with sub-queries, which is more performance (thanks Nick and Blake for testing against your larger DB!)
* Update peewee index migration description
* Add on_conflict_ignore, replacing the try/catch/pass on IntegrityError
* Add a testcase for validating that on_conflict_ignore bypasses what was formerly an IntegrityError
* Change testcase to clarify that it covers the peewee behavior of on_conflict_ignore
---------
Co-authored-by: Greg <{ID}+{username}@users.noreply.github.com>
* add a notice registry and System Health tab
Problems Frigate detects on its own (ffmpeg crash loops, stuck detectors, failed model downloads, recordings deleted before their retention period) only ever existed as log lines. This adds a `NoticeRegistry` in the main process backed by two tables, an `update_notice` IPC topic so producers in other processes can reach it through the dispatcher, an admin-only API and websocket topic, and a Health tab that lists them. Kinds declare their own mode, severity, and category in one place: state notices are resolved by their producer, event notices are dismissed by the user.
* treat a prerelease as behind its final release
* fix notices clearing early
* rename menu items and update docs
* don't resolve the update notice on a failed version lookup
isLoading and isBuffering only clear on playback progress, and scrubbing holds the player paused, so a source rebuild during a timeline selection left them set with nothing able to clear them. Cancelling made them visible as a spinner over an already-loaded frame, which stayed until the next manual seek. Leaving a scrub now clears them when the source is loaded and the element holds a frame.
* Refactor hardware stats to have consolidated ffmpeg, detector, and enrichments running.
* Cleanup hardware access that is not passed into the container
* Remove network stats from hardware refactor
`after` and `before` defaulted to `datetime.now()` in the function signature, so they were evaluated once at import. Requests that omitted them got a window ending at process start. They now resolve in the handler.
* Support read-only rootfs with self-signed certs in /config/tls
* Support read-only rootfs in s6 and pre-compile bytecode
* Assert read-only rootfs support in CI
* Document hardened read-only deployment
* keep certsync's cert selection identical to nginx's
* note the uid trade-off in user: mode
* fail fast when EXTRA_GROUPS or a missing media volume meets read_only
* keep nosuid and nodev on the /run tmpfs
* support read_only in the default mode
* don't take go2rtc down when the homekit file isn't writable
* lead with the hardware consequence of switching to user:
* refuse to write TLS material through a symlink as root
* note that memryx writes models to the root filesystem
* certsync watches whichever cert path nginx loaded
* Run the frigate service as the frigate user
* Run go2rtc as its own restricted user
* Run nginx as the frigate user with writable state in /tmp/nginx
* Disable bandwidth stats gracefully when not running as root
* Hand TensorRT model cache ownership to the runtime user
* Document non-root operation and per-hardware device access
* Create /media/frigate after the ownership sweep
* Assert non-root services, JWT migration, and escape hatch in CI
* only write the sweep sentinel when a media volume is mounted
* tolerate homekit config chown failures in the go2rtc run script
* chown the s6 log pipe so non-root nginx can reopen /dev/stdout
* set HOME to /config for non-root services
* run smoke nginx -t and the write probe as the runtime user
* re-own the nginx shm cache on service restart
* discard stdout for the unprivileged smoke nginx -t
* unwrap hard-wrapped prose in the installation docs
* report progress during the ownership sweep
* document EXTRA_GROUPS as the only device access path for dropped services
* expand the non-root device access docs with diagnosis steps and udev rules
* document network storage ownership and the remaining detector hardware
* skip lost+found during the ownership sweep
* hand /tmp/cache to the runtime user before services start
* make bundled models readable by the runtime user
* reload nginx by signaling the master instead of parsing its config as root
* harden root writes into unprivileged-owned paths
Restrict the sweep sentinel to a mount at or below /media/frigate so a
parent /media mount cannot bless a later-shadowed volume. Rebuild
/tmp/nginx root-owned each start so root's cp and tempio writes cannot
follow a symlink an unprivileged nginx planted in the previous run.
* collapse the duplicated sentinel comment
* add a service-runs-as-root helper for granular root services
* validate FRIGATE_ROOT_SERVICES and fail fast on unknown names
* let services listed in FRIGATE_ROOT_SERVICES skip the privilege drop
* record the root-services mode in the sentinel and sweep small trees each boot
* cache the runtime ids in the ownership helper
* chown recordings, previews, and exports to the runtime user at create
* chown the database files after init
* recommend FRIGATE_ROOT_SERVICES in the bandwidth stats warning
* assert granular root services in CI
* document FRIGATE_ROOT_SERVICES
* own every directory level created for a recording segment
* clear the cached runtime ids when ownership tests finish
* skip missing media paths in the per-boot ownership sweep
* clarify granular root services docs
* clean up
* install acl for device access grants
* grant runtime users access to mapped device nodes at boot
* assert device access grants in CI
* document automatic device access grants
* stop telling users device access needs host side setup
* clarify the non-root docs
* link the migration script to the repo
* group the manual device setup under one section
* harden against symlink attacks
/config is owned by the unprivileged runtime user after the ownership sweep, so root operations on files there could be redirected by a planted symlink.
- go2rtc HomeKit setup: replace the root yq/jq normalization and chown with an O_NOFOLLOW helper (prepare_homekit.py), so a symlink at go2rtc_homekit.yml can't redirect a root write or chown onto another file
- go2rtc binary override: ignore /config/go2rtc whenever the service runs as root, so a planted binary can't exec as root under FRIGATE_ROOT_SERVICES
- sweep sentinel: read and write it through safe-sentinel, which trusts only a root-owned regular file and never follows a symlink, so it can't be forged to skip the migration or symlinked to clobber a root file
- ownership sweep: chown with -execdir so a parent directory swapped for a symlink mid-walk can't redirect the chown out of the volume
- validate inputs: restrict DEVICE_ACL_PATHS to /dev, require nonzero numeric EXTRA_GROUPS, and reject PUID/PGID that collide with the go2rtc ids
- docs: correct the TLS key ownership note to match what actually happens
* tweak docs
* stop the ownership sweep chasing entries other mechanisms own
* keep custom binaries out of root services only under granular root
The `v1.x` line is the same muxed fMP4 code as Kaltura's 1.31 with fixes backported, so the mapping JSON, manifest routes, and ffmpeg consumers are unchanged. The `MAX_CLIPS` patch applies as-is and the HEVC workaround is rewritten for the fork's reformatted source.
`vod_hls_version 6` is now explicit because the fork replaced Kaltura's automatic version calculation with a directive that defaults to 4 and only warns when fmp4 needs 6, so playlists were being stamped `EXT-X-VERSION:4` while carrying `EXT-X-MAP`. The `error_page 502 =404` hack is gone: https://github.com/kaltura/nginx-vod-module/issues/468 is a `vod_mode remote` bug and we're `mapped`, so those 502s were really `_vod_response` returning 404 upstream. The fork maps that through now, and the hack was also turning real 5xx into "no recordings".
* use the camera's model for the lpr vehicle check
`FrigateConfig.model` became `models[]` in the detector refactor, so this didn't compile on 0.19. Each camera has its own detector/model pair now, so the check resolves the camera's scene with `getModelForCamera` instead of looking at every model.
* use camera config model
* fix export test
* add onboarding wizard for new users
* resolve hwaccel per camera and clarify recording retention
The hwaccel step listed every preset Frigate ships, so an Intel box was offered Raspberry Pi and Rockchip decoding, and the codec specific presets (`preset-intel-qsv-h264` vs `-h265`) were offered as global values that break as soon as two cameras use different codecs. `/hardware/hwaccel` now returns the decoding families the probed hardware can actually use, each carrying a preset per codec, and the wizard resolves the family against the detect stream codec the camera wizard already probed: one global `ffmpeg.hwaccel_args` when every camera agrees, per-camera `cameras.<name>.ffmpeg.hwaccel_args` when they don't. The global stays on `auto` in that case so cameras added later still resolve at startup. A gen13+ Intel machine keeps its QuickSync recommendation with mixed h264 and h265 cameras instead of dropping to vaapi.
The recording step's "Days to retain recordings" only wrote alert and detection retention, and the storage estimate under it assumed continuous recording. It now asks what to record in plain language, writes `record.continuous.days` to match, shows the estimate only for continuous, and drops the spinner arrows on the number input.
* clean up
* add light/dark mode icon switcher
* use yml as default config file extension when not found
* i18n tweaks
* gate the setup wizard on cameras instead of a config key
* render setup wizard steps by key
* share the setup wizard e2e helpers and mock users
* add an account step to the setup wizard
* add setup wizard account step e2e coverage
* cover the account step's restart behavior
* button consistency
* fix test
* docs
* fixes
* fix inconsistent export download filenames
Zip entries in a case download were named from `Export.name`, the friendly display name, while an individual download uses the file name on disk. The two have always been formatted differently, so one export came out as `front_door_20260823_020615-20260823_020734_abc123.mp4` on its own and `front door 2026-08-23 020615 2026-08-23 020734.mp4` inside a zip. Zip entries now use the on-disk file name, and renaming an export renames its file, so there's only one name to download under. The rename is blocked while ffmpeg still holds the file.
* cap filename length and catch duplicate names
* fix export rename and stop blocking the event loop
* move the rename rollback off the event loop
* no awaits
* pin genai review frames to the main stream
* retain previews as long as either stream has recordings
* watch sub stream recording health separately from main
* reject record_sub on the same input as record and document the role
* derive recording paths from the cache segment timestamp
Recording paths carry one second of resolution, but since sub stream recording start times are resolved to fractional wall clock, anchored to the cache file mtime and chained to the previous segment's end. A stream cutting segments faster than once a second resolves consecutive segments into the same second, so two rows collide on the unique path index and the batch insert fails. The cache segment name is unique per camera stream and second by construction because ffmpeg names segments with strftime, so the recording path is now built from that timestamp while the row keeps the resolved start time. This also restores the path semantics from before sub stream recording, when start times came straight from the cache filename.
Nothing derives times from recording paths: playback offsets, stream switching, and export all use the row's start time, which is unchanged, and the recordings sync matches files by exact path string.
* keep the rest of a recording batch when one row conflicts
* only publish record_sub status when a sub stream is configured
* don't shadow camera_cfg when publishing empty cache streams
* back off restarts when a recording stream goes stale
* give the shared sub stream grace on any capture thread reset
* include segment details in recording discard warnings
* Create frigate and go2rtc runtime users in the image
* Add single fix-ownership helper for volume permission migration
* Add init-usermod oneshot for PUID and PGID remapping
* Chown newly created runtime directories to the frigate user
* Run sentinel-guarded ownership sweep during prepare
* Add host-side volume permission migration script
* Guard log directory ownership for user-mode startup
* Fall back to plain s6-log when running without root
* Assert PUID remapping and sweep sentinel in CI smoke test
* Skip the ownership sweep in the devcontainer
* Pin FRIGATE_RUN_AS_ROOT in ownership tests
* Do not record the sweep as complete when a chown failed
* Validate PUID and PGID in the migration script
* Treat a failed ownership scan as an incomplete sweep
* Reject PUID and PGID of 0 during remapping
* Handle symlinks, dry runs, and sentinel write failures in the sweep
* Treat an absent sweep root as an incomplete sweep
* Verify s6-overlay downloads against pinned checksums
* Verify go2rtc download against pinned checksums
The v1.9.14 release publishes no checksums file, just the bare per-platform binaries, so these digests come from a one-time fetch rather than upstream. That pins the artifact against later substitution, which is the realistic threat for a version we stay on for months, but it does not verify the original download. The stage moves from `ADD --link` to a script because `ADD --checksum` can't express an architecture-dependent URL.
* Verify main image downloads against pinned checksums
Covers everything the main image downloads on the default path: tempio, the hailort runtime tarball and wheel, the six ffmpeg builds, the libedgetpu deb, and the thirteen Intel driver debs. The hailort tarball was streamed straight into `tar`, which can't be verified before extraction, so it downloads to `/tmp` first. The three ffmpeg blocks per arch collapse into one `install_ffmpeg` helper since they only differed by URL and install dir, and the Intel debs go through a `fetch_intel_deb` helper for the same reason.
The Intel debs are the ones that mattered most here. They're installed as root with `dpkg` on the default amd64 path and had no verification at all. compute-runtime publishes a `ww<week>.sum` asset with every release and npu-driver published `checksum.sha256` on v1.19.0, so those eight digests came from upstream rather than from us. intel-graphics-compiler and level-zero publish none, so those five and everything else here come from a one-time fetch, which pins the artifact against later substitution but doesn't verify the original download. The comment above the map says which is which and how to refresh them, since npu-driver has stopped publishing sums since v1.19.0 and that provenance won't survive the next bump.
Still unpinned: `get-pip.py`, which is a rolling URL where a digest would just break the build on pypa's next edit, and the per-variant artifacts for Axera, Synaptics, and Jetson. apt repositories are out of scope since apt already verifies signatures.
* Restrict generated TLS key permissions
OpenSSL 3.x already writes the key at 600 on its own, so this pins the guarantee rather than fixing an observed leak: the mode no longer depends on the openssl version or the umask the service happens to start with. Only the generated pair is touched. User-mounted certs take the other branch and are never chmod'd, which matters when they're mounted read-only.
* Add security headers and server_tokens off
Adds `X-Content-Type-Options: nosniff` and `Referrer-Policy: strict-origin-when-cross-origin`, and turns off nginx version disclosure.
No `X-Frame-Options` and no CSP `frame-ancestors`. HA's Webpage card and iframe panels frame Frigate's own address cross-origin, and either header would break them silently with nothing in Frigate's logs to explain it. Ingress is same-origin and would survive `SAMEORIGIN`, but Frigate can't tell the two apart from inside the container. `security_headers.conf` is a plain file in the image rather than a generated one, so anyone who does want framing restrictions can bind-mount it.
`add_header` doesn't inherit into a block that declares its own, so the include goes in per block, all nine of them, including the four nested static-asset locations that serve the JS bundles. Those are the ones nosniff actually matters for.
The run script now reads `get_nginx_settings.py` once into a variable instead of shelling out per template. That script imports the frigate config machinery, which is noticeable on an SBC.
Not fixed here: `listen.conf` is included at server level and carries `Strict-Transport-Security`, so those same nine blocks already drop HSTS under TLS today. Folding it into this file would change existing TLS behavior on nine paths, so it needs its own PR.
* Restrict go2rtc config file permissions
* Log failed login attempts with source address
Failed logins returned a bare 401 and left nothing behind, so credential stuffing was invisible unless you were already watching nginx access logs. Both failure branches now log a warning with the attempted username and the client address.
The address comes from `get_remote_addr()`, the same helper the login rate limiter keys on, so the two agree on who the client is and the trusted-proxy handling is consistent. Logging a raw `x-forwarded-for` instead would let an attacker forge the source address in the very log line meant to catch them.
The response is unchanged and identical either way. Which factor failed is only visible in the log, never to the client, and the password is never logged.
* Recommend least-privilege container options in install docs
The compose generator pushed `privileged: true` into every file it produced, no matter what hardware you picked, and it's the default tab on the install page so it's what most people copy. It now emits `security_opt: no-new-privileges:true` instead, and only adds `privileged: true` for hardware that actually needs it, with the reason inline. MemryX is the only one today, since it needs to reach the max-manager. Rockchip and Synaptics only want privileged during initial setup and their documented end state is device mappings, so neither gets it.
`no-new-privileges` merges into the same `security_opt` block as any device-specific entries, so Rockchip still gets its `apparmor=unconfined` and `systempaths=unconfined` without a duplicate key.
The static example now has `privileged` commented out, and there's a short section on the options worth adding, with a note that `cap_drop: ALL` breaks `telemetry.stats.network_bandwidth` since nethogs needs NET_ADMIN/NET_RAW.
* Add amd64 container smoke test to CI
Boots the built amd64 image against a minimal config and asserts the two security headers, that the Server header no longer carries a version, that no frame-ancestors is present, that nginx accepts its own config, and the two file modes. This is also the harness the rest of the hardening work extends.
The two negative assertions are written as `if grep; then exit 1; fi` rather than `! grep`. Bash exempts a negated command from `set -e`, so the `!` form would have passed even with the version and frame-ancestors both present, which is the opposite of what a regression net is for.
* improve keyframes messages
* don't pad the labelmap with unknown
`load_labels()` prefilled 91 `unknown` entries before reading the label file, so any model with fewer than 91 classes kept that padding in `merged_labelmap` and `unknown` showed up as a selectable object type in the objects settings UI. The padding only existed so `RemoteObjectDetector.detect` could index the labelmap without a KeyError, and it didn't even cover the empty-file case or Frigate+, which never had a prefill. Both lookups now skip class ids the labelmap doesn't name and warn once per id.
* add secrets.yaml and merge substitution sources by precedence
FRIGATE_ENV_VARS was built once at import from container env and /run/secrets, and the environment_vars validator overwrote it unconditionally, so the block beat the deployment and nothing could be re-read. Sources are now separate dicts merged lowest to highest (environment_vars, secrets.yaml, container env, credentials directory), re-read at the top of every parse, and a collision warns once naming the winner. An undefined {FRIGATE_*} raises a ValueError subclass so pydantic reports the field instead of a KeyError traceback.
* use the shared substitution namespace in go2rtc config
The generator rebuilt the namespace itself from os.environ and a hardcoded /run/secrets, so it never saw environment_vars or CREDENTIALS_DIRECTORY, and str.format made any stray brace fatal. It now installs the FRIGATE_ names from environment_vars and substitutes streams the same way every other field does.
* read the exec override from an import time snapshot
environment_vars is exported into os.environ, and is_go2rtc_arbitrary_exec_allowed read os.environ live, so the config file could enable exec sources. Snapshot the variable at import, which runs before any config is loaded.
* docs
* clarify docs
ffmpeg's stderr was piped but never read, so recording segments that generate more than 64 KB of ffmpeg warnings blocked ffmpeg mid-write, stranding the streaming thread and its anyio threadpool token for good. Enough of those and every sync endpoint stops responding until restart. The trigger is how noisy the segments are, not how long the clip is.
Send stderr to a temp file instead, and guarantee ffmpeg teardown and playlist cleanup on every exit path, including client disconnect.
Also fixes two bugs the deadlock hid: the failure branch was unreachable because returncode is None mid-loop, so the playlist file leaked and ffmpeg's logs were never reported. Playlist files now get a unique name so concurrent requests for one range cannot delete each other's input.
Extracts the terminate helper motion search already had into frigate/util/ffmpeg.py, now shared by both streaming call sites.
* add import/export for camera group layouts and streaming settings
Camera group layouts and per-camera streaming settings are stored in the browser's IndexedDB, so they are tied to a single browser on a single device. Users with more than one device have to rebuild every group layout and re-pick every camera's stream settings by hand, and clearing browser data loses the work.
Add a Backup & Restore card to Settings > UI Settings that exports these settings to a JSON file and imports that file on another device. Import shows a confirmation dialog with per-section counts, switches for layouts, streaming settings, and UI preferences, and warnings about camera groups or cameras in the file that are not on this server.
Server-side storage is deliberately avoided. These are per-device presentation settings: a layout arranged for a desktop is wrong on a tablet, and continuous full-resolution streams that are free on a wired LAN are not on a phone. An explicit file moves settings only when the user chooses to move them.
Implementation notes:
- web/src/utils/uiSettingsTransfer.ts owns a registry of transferable IndexedDB keys. Each entry records whether the key is user-namespaced, matching which persistence hook wrote it, plus a zod schema for its value.
- Only registry-known keys are ever written, and only when their value passes that schema. The file format deliberately lets unknown keys survive parsing, so this filter is what prevents a hand-edited file from writing arbitrary storage keys or out-of-range values.
- Export falls back to the legacy un-namespaced key, because the username migration runs lazily on first mount of each owning hook.
- Streaming settings merge per group rather than replacing the whole map, so groups configured only on the receiving device survive.
- Import writes storage and then reloads, because useUserPersistence reads a key only on mount and StreamingSettingsProvider would otherwise write its stale in-memory state back over the import.
- playbackBandwidthEstimate, frigate-search-history, and live-layout are excluded: the first two are measurements and user data rather than preferences, and live-layout's default is derived from the device.
* merge imported streaming settings per camera instead of per group
Allow audio classes to be grouped under a shared configured label.
Keep audio overrides separate from object labels and retain only the highest-scoring grouped detection.
Refs #23967
* refactor mqtt so that Frigate owns the transport lifecycle instead of delegating it to paho
* release the shutdown barrier on worker crash and replay retained publishes the broker never acked
* collapse in-flight retained values by topic and release the shutdown barrier from a finally
* replay the outage buffer before the publish queue so newer values are not reverted
* serve a segment startup ladder so seeks begin playing sooner
nginx-vod was handed one 10s segment per recording file, so every playlist start had to download and decode a full segment before the first frame. Declare real keyframe data per clip and let nginx cut short leading segments from it.
- add vod_bootstrap_segment_durations 1000/2000/4000 so each playlist starts with 1s/2s/4s segments before settling at 10s
- emit real clip-relative keyFrameDurations (plus firstKeyFrameOffset when nonzero) from the recording keyframe index; rows without an index keep the whole-clip declaration, the only safe cut without keyframe knowledge
- drop the manifest's segment_duration field, which was always inert: nginx-vod parses only camelCase segmentDuration
- rebuild the player source at the seek target, quantized to a 10s grid, so the ladder applies to every seek and seek URLs stay repeatable for nginx's mapping and response caches
- route the seek model, in-range checks, and the stale-report guard through the source window rather than the chunk range
- bridge repositioning seeks (>2s from the last played timestamp) through the preview player and hold the release anchor one commit, so neither path paints a stale frame
- clear a pending loading timer before replacing it; an orphaned timer escaped onPlaying's clearTimeout and flashed loading mid-playback
* keep recordings queries on their indexes
Several recordings queries degraded into full scans or large sorts on big databases: the planner ignored index order, or the query shape gave it nothing tight to seek on. Reshape them into bounded seeks and add the composite index the per-stream lookups need.
- index recordings on (camera, stream_type, start_time DESC) and drop the (camera, stream_type) index it supersedes
- walk the recordings summary day by day with EXISTS probes and per-camera MIN/MAX seeks, skipping ahead over empty gaps instead of bucketing every row for the requested cameras
- run the summary endpoint on the event loop rather than the threadpool
- bound the unavailable-recordings query by start_time per camera and merge the results in Python
- bound the expire query's start_time so it seeks the retention window instead of scanning a camera's whole history
- enumerate deleted cameras with one index seek each rather than a camera NOT IN (...) scan
- compute bandwidth with segment_size filtered in a CASE projection; as a WHERE predicate it baited the planner into the (camera, segment_size) index plus a full sort of the camera's history
- fall back to a 1000-segment window when the recent 100 are all zero-size, so an ingest glitch doesn't report zero bandwidth
- limit the needs_refresh count instead of counting every segment
- cover sub-only and sparse calendar days, midnight-spanning day attribution, multi-camera gap merging, deleted-camera expiry, and zero-size segment runs
* fix mypy
* add ptz controls to camera via wizard when onvif has already been probed
* i18n
* add e2e test
* backend add and remove subscriber
* tweaks
* turn on switch by default if pan and/or tilt capability is available
* fix test
* add sub stream recording with adaptive quality playback
Optionally record a second, lower bitrate stream alongside the main
recording stream via a `record_sub` input role and `record.sub` config block, with its own retention windows.
Recordings rows now carry the stream type plus the media details needed to serve both streams from one manifest: video codec, audio presence, audio codec and rate, and a record-time keyframe index.
Playback resolves coverage across both streams and merges them into a single VOD sequence, falling back to a discontinuity manifest with per-clip init segments when the media signatures differ. The player exposes a quality selector, and an auto governor picks the stream from stall time, bandwidth, codec support, and the save-data hint.
* fix tests and i18n
