SDK changelog
What each version of the SDK and capture-server brings, the breaking changes and the compatibility policy of the 0.x line.
The latest versions of both packages, with what each brings. Each package ships its changelog (CHANGELOG.md) from capture 0.36.0 and capture-server 0.14.0 on; earlier versions come from a copy of the previous changelog that DevLab keeps. The notes are quoted as the package wrote them, so some may be in Spanish. This page is generated with npm run docs:gen; do not edit it by hand.
@neurologyca/capture
Section titled “@neurologyca/capture”0.45.0 · 2026-11-16
Section titled “0.45.0 · 2026-11-16”A proxy on another origin. No breaking changes.
- Added: Option
credentials('omit','same-origin'or'include') oncreateProtocolClientandcreateRecordingClient: thecredentialsmode of the client’s requests to the Engine or your proxy, including the Engine’s event stream and the streamed interpretation (not the startup report, which always carries cookies). - Added:
urlTokenProvider(tokenUrl, { credentials }): the same option for the token request. - Fixed: At the close of a capture, a
POST /analyzeasked again after a202 ANALYSIS_IN_PROGRESSnow gets as its own timeout what is left of the in-progress wait (analyzeTimeoutMscounted from the first202), never less than 10 000 ms. - Fixed: The
interpretation.sse-timeoutandanalysis.regeneratedwarnings that follow a202 ANALYSIS_IN_PROGRESSnow count the in-progress wait in the time they say the Engine’s announcement did not arrive within.
0.44.0 · 2026-11-16
Section titled “0.44.0 · 2026-11-16”An analysis in progress is said, not lost. No breaking changes.
- Added:
client.analyze()(oncreateProtocolClientandcreateRecordingClient) resolves to the newAnalyzeAnswertype, which addsinProgressandretryAfterMstointerpretation,fromCacheandraw. - Added: Warning
analysis.in-progress: at the close of a capture,POST /analyzeanswered202 ANALYSIS_IN_PROGRESS. - Added: Error
E_ANALYSIS_IN_PROGRESS(recoverable: true): at the close of a capture, the Engine was still generating the interpretation when the wait ran out. - Changed: At the close of a capture, a
202 ANALYSIS_IN_PROGRESSfromPOST /analyzeis waited out: the capture waits what the Engine asks (30 000 ms when it does not say, never less than 1 000 ms), still listening for the Engine’s end-of-interpretation event on the stream withinterpretation: 'engine', and asks again, withinanalyzeTimeoutMscounted from the first202.
0.43.0 · 2026-11-13
Section titled “0.43.0 · 2026-11-13”Models kept between captures. Breaking
- Added:
keepPreparedoption ofcreateProtocolClient(off by default; live only). - Added: With
keepPrepared, aprepare()that asks only for what is already kept (the same parts or fewer) resolves without loading anything: its result is a new one (a newreport.startupId, its owncache,totalMsand device diagnosis), its stages arrive already closed, one per part, and itsStartupReportcarrieskept: true(schema: 2). - Added: With
keepPrepared, thepreparedblock of theStartupReportof a capture or a baseline that was lent models kept from an earlier baseline or capture carriesprepared.origin({ face, speech }, each'prepare','previous-session'ornull; typePreparedOrigin), whether it used them or not, and the report isschema: 2. - Added: With
keepPrepared, a kept part that kodec leaves unusable while it waits for the next capture (its engine died, its engine kept failing, it did not settle in time, it stopped producing frames, it lost its GPU context, or it failed to load again) raises aprepare.releasedwarning with the new reason'unusable'and itspart, and only that part is released; with nothing left alive, everything kept is released with one such warning: itspartis the first dead part (the face before the speech) and its message names every dead part with its reason. - Added:
prepare.face-lostwarning (withpart: 'face'andfields: null): the prepared face models rejected their first frames, or lost their GPU context (the message says which), and kodec loads them again, once. - Added:
face.context-lostwarning (withpart: 'face'andfields: null): the face analysis a capture or a baseline loaded itself (not a prepared one) lost its GPU context (a driver reset, resuming from sleep, a GPU process crash) and kodec loads it again, once; there are no frames while it reloads. - Added: kodec reloads the face on the main thread, so its
capture-stalledwithcause: 'main-thread-blocked'for a gap that falls within that reload is the reload itself: it is not acodec.errorand is not counted inSessionReport.codecErrors, in a capture or a baseline. - Added: A bound on that reload once the capture has its first frame: if the face analysis is not back within
startup.timeouts.vision.totalMs(120 000 ms when that cap isnull), the capture fails exactly as when kodec cannot reload the face: acodec.errorwarning withcapture-failed, counted incodecErrors, and kodec stopped. - Added:
PreparedBlock.faceReloaded:truewhen kodec had to load the prepared face again, otherwise absent; the face still reads'adopted', and the report isschema: 2. - Added: Startup message key
first-frame.face-reload-failed(es/en text inSTARTUP_MESSAGES): the face analysis stopped working as the capture started and could not be set up again (afterprepare.face-lostorface.context-lost). - Changed: kodec
5.0.0-rc.14 - Fixed: Without a transcription worker (a Content-Security-Policy that blocks workers, or a browser where the worker does not start), stopping a capture or releasing the models now frees the speech model’s GPU memory, after any inference in progress finishes; before, it was never freed, and GPU memory grew with every capture (kodec
5.0.0-rc.14). - Fixed:
SessionReport.transcription.workerBlockedByCspnow readstruewhen the page’s Content-Security-Policy blocks the transcription worker in browsers that report the block only after the worker is created; before, it readfalsethere (kodec5.0.0-rc.12). - Fixed: When a capture stops and stopping its frame pump throws, the camera and the analysis engine are still stopped; before, they could keep running.
- Fixed: A webcam whose video track ends later in the startup now fails the startup at once with the error of a webcam that stops before the models load (
CaptureStartError,stage: 'source',reason: 'error', «the camera stopped sending video while the capture was starting: check that no other app is using it and try again»), with noE_CAPTURE_LOST, noinvalidatedstatus and no first-frame timeout. - Breaking (types):
StartupReport.schemais now1 | 2. - Breaking (types):
PreparedBlock.prepareIdis nowstring | null, andPreparedBlockgains the optionalorigin. - Breaking (types):
ProtocolWarningEvent.reasongains'unusable'(onprepare.released). - Breaking (types):
SessionReport.transcriptionLoad.reasongains'needs-worker'(withstage: 'stt-session'): transcription was allowed only byallowWasmWorkerand its Worker did not start, so it was not enabled. - Breaking (types):
StartupMessageKeygains'first-frame.face-reload-failed', andSTARTUP_MESSAGE_KEYSlists twenty keys: aRecord<StartupMessageKey, string>of your own needs its text.
0.42.1 · 2026-11-12
Section titled “0.42.1 · 2026-11-12”Startup messages that fit the source. Breaking
- Changed: The five
face-model.*startup messages (face-model.running,face-model.cached,face-model.stalled,face-model.timeout,face-model.error) now speak of getting face analysis ready instead of the camera, inesanden: the face model loads the same with a webcam, astream, a<video>or no source at all (a preparation). - Added:
first-frame.timeout-media: the message of a first-frame timeout when the capture or baseline source is your ownstreamor<video>(«No face can be seen in the video…», inesanden), and in the replay of a recording. - Changed (messages for developers): The
messageof a first-frameCaptureStartErrorsays «of starting the source» instead of «of starting the camera». - Breaking (types):
StartupMessageKeygains'first-frame.timeout-media': an exhaustiveRecord<StartupMessageKey, string>(for instance a full set of your own texts) must add it.
@neurologyca/capture-server
Section titled “@neurologyca/capture-server”0.20.1 · 2026-11-16
Section titled “0.20.1 · 2026-11-16”Retry-After readable from another origin. No breaking changes.
- Fixed:
createEngineProxyexposesRetry-Afterto the browser
0.20.0 · 2026-11-16
Section titled “0.20.0 · 2026-11-16”An analysis in progress, and a kit for Measurement. No breaking changes.
- Added:
AnalyzeResult.inProgressandAnalyzeResult.retryAfterMs - Added:
ProtocolSummary.interpreted - Added:
TenantKit.kind - Changed:
buildTenantKitgives a kit for Measurement.
0.19.0 · 2026-11-13
Section titled “0.19.0 · 2026-11-13”The startup report receiver validates schema 2. No breaking changes.
- Changed: The startup report receiver now validates schema 2 reports.
0.18.0 · 2026-11-12
Section titled “0.18.0 · 2026-11-12”Startup report receiver. No breaking changes.
- Added:
createStartupReportReceiver - Added:
localDiskStartupReports({ dir, maxFileBytes })
0.14.3 · 2026-10-29
Section titled “0.14.3 · 2026-10-29”The protocol group’s target HCI keeps the Engine’s field name. Breaking
- Breaking:
ProtocolSummary.group.targetHciIdis nowgroup.targetConstellationHciId
DevLab
Section titled “DevLab”- 2026-11-16: The documentation, the kit and the starters describe
@neurologyca/capture0.45.0 and@neurologyca/capture-server0.20.1. - 2026-10-12: The documentation describes
@neurologyca/capture0.27.0 and@neurologyca/capture-server0.9.0: the SDK first, its two processing modes, and the HTTP API as a complement.
Compatibility and support
Section titled “Compatibility and support”Both packages follow semver on a 0.x line: a minor breaking change can land in a patch release, and then it is flagged Breaking. From 0.4.0 on, every version speaks Engine 2 only. 1.0 is kept for production. See Release phases.