The received-files lookup above a rendered folder now treats a record path
that exists but is not a readable file (a folder, a dangling link, any
other kind of file) as an unreadable record, the same answer the folder's
own record gives. Only a missing record means nothing was received.
A listed file now counts as inside the rendered subfolder when one of its
parent folders is the same folder on disk, so naming the subfolder in
another letter case on a case-insensitive disk keeps its received files.
* fix(parsers): helper lookups skip the project above the working folder, not only the folder
* fix(parsers): compare a missing PATH folder by its nearest real parent, so Windows short names match
* fix(engine): extraction-time video downloads get the not-media check
A video src that is still remote at frame extraction is downloaded again and
then probed. That copy now gets the checks other media downloads get before
anything reads it as media:
- the not-media sniff, so an HTML, XML or JSON body (including a DASH
manifest) is refused with the existing NotMediaPayloadError, classified as
invalid_media;
- on a render service (PRODUCER_RESTRICT_PAGE_NETWORK), the same
-format_whitelist an https input gets, so a playlist, manifest or concat
list fails to open whatever its bytes say, and no file it names is read.
Real mp4, mov and webm downloads extract the same frames as before.
* fix(engine): the download format check stops when the render is cancelled
* fix(engine): the not-media sniff keeps mp4 files whose box size reads as text
An ISO BMFF file opens with a 4-byte box size and a box type. A size of 60
is the byte `<`, so after the leading NULs were skipped a valid mp4 with a
60-byte ftyp was refused as a text document. The sniff now recognises an
mp4/mov box header (size 0, 1 or at least 8, then ftyp, styp, moov, moof,
mdat, free, skip, wide, sidx or pdin) before it skips NULs. HTML, XML,
JSON and playlist refusals are unchanged.
* fix(engine): the mp4 exception in the not-media sniff needs a box that fits the file
The sniff only skips its document check for an mp4/mov opening when the
first box is ftyp or styp, its declared size (including a 64-bit size) fits
the file length, and the next box header, when the sniff window holds it, is
a known media box that also fits. A document that only spells a box header
falls through to the document and playlist checks as before.
* fix(engine): the not-media sniff keeps mp4 files with leading padding or extension boxes
An mp4 or QuickTime file may open with free, skip, wide, uuid, pdin, moov,
mdat or moof as well as ftyp or styp, and may carry any box type after it.
The sniff now accepts any of those first boxes, then walks every box header
the sniff window holds: each needs a printable four-character type and a
size that fits the file. A document that only spells a box header still
falls through to the document and playlist checks.
The outside-project media check skipped any src starting with data:, blob:,
about: or http(s): in any letter case, while the readers resolve everything
but exact http:// and https:// URLs as a local path. A src such as
data:x/../../outside/clip.mp4 passed the check and the readers opened the
file it named.
The check and the media type preflight now use the readers' remote test,
isHttpUrl, and check every other src with a shared containment helper,
isInsideAssetRoots. Frame extraction, the audio mix, and the HDR and video
sound probes also refuse a local source outside the project, compiled and
download folders on their own when outside files are not allowed.
The CLI static project server and the producer and engine file servers bind
127.0.0.1, but answered any Host header. They now refuse a request whose Host
is not 127.0.0.1, localhost or [::1] at their own port (or the in-process
origin the CLI hands out when listen is refused), and their responses carry
frame-ancestors 'self' and X-Frame-Options: SAMEORIGIN so another site cannot
frame the served page.
The notice a render prints for a received project, and the left-out warnings and
outside-files list that publish, feedback and cloud render print, now show control
characters from references, file names and paths as visible escapes instead of
sending them to the terminal.
While it names the references a received project makes outside itself, the render
check no longer looks at a path on another computer's share or a Windows device
path. It names such a reference in the notice without reading it. A Windows device
prefix such as \\?\ is no longer cut short as if its ? started a query.
A HEYGEN_API_KEY or HYPERFRAMES_API_KEY that a project .env sets now comes
after every credential the person saved or exported. The CLI resolver, the
media environment and the media-use skill use it only when there is no saved
login, no saved API key and no key in the shell. When one is used, a single
stderr line names the .env file.
An expired, foreign-client, unreadable or refused login now raises the login
error instead of switching to a project .env key, whether or not the folder
has a received-files record. A key exported in the shell, or set by the app
that launched the CLI, keeps its current place, and the host gateway path is
unchanged.
* fix(parsers): helper programs never come from the project a subfolder sits in
Run from a subfolder, the trusted PATH check skipped only folders inside the
working directory, so the project's own node_modules/.bin, which npm run puts
on PATH from every ancestor, still supplied helpers. The check now skips every
folder inside the outermost ancestor that marks a project (.git, package.json,
node_modules, hyperframes.json), below the home folder.
A CLI test also runs a real bare-name spawn of an executable planted in the
project folder: it runs under the raw PATH and is not found after the PATH trim.
* fix(parsers): keep the trusted lookup's comment share at main's level
* fix(cli): the audio caches never write or read through links in .hyperframes
The read-audio cache, its stems, mix and descriptions, and the audio-kind cache
now use only real folders under the project's .hyperframes. A link or a file in
place of a cache folder stops the write with a clear message. Every cache file is
written to an exclusive temp file with a random name and renamed into place, so a
link at a temp or final name is replaced, never written through. ffmpeg writes the
voice-free mix into a private staging folder. Stem downloads use the new
downloadToFile replaceLink option. Cached files are read only when they are plain
files, and a stem list may only name files in its own folder.
* fix(cli): the kind cache reads no entry through a linked cache folder
readKindCache now checks that .hyperframes and audio-kind are real folders before
it reads an entry, as writeKindCache already does before it writes one.
* test(cli): read the replaced download before checking its type and mode
The replaceLink test now reads the file's bytes and lstat result once each, so
no read follows a check on the same path.
A download whose response declares a Content-Length re-arms its attempt
timeout on every chunk, so a server that trickles just above the throughput
floor was never cut off by time. Parallel downloads also had no shared limit
on the disk they could fill.
- Each download attempt now has a deadline that incoming data never moves:
30 minutes by default, HYPERFRAMES_MAX_DOWNLOAD_MS to change it. Passing it
fails the download without a local retry.
- The default per-file cap drops from 20 GiB to 8 GiB
(HYPERFRAMES_MAX_DOWNLOAD_BYTES still overrides it).
- Downloads keep 512 MiB free on their volume
(HYPERFRAMES_DOWNLOAD_MIN_FREE_BYTES, 0 turns it off). A declared size is
claimed before the body is read and counted against every parallel
download on the same device, and free space is measured again every 4 MiB
as bytes arrive. Free space that cannot be measured does not block.
- docs/packages/engine.mdx lists the download limits and their variables.
* feat(studio): hosts can seal the preview frames NLEPreview draws
* refactor(studio): Player builds its frame src in a helper, keeping its complexity at main's
* fix(cli): snapshot shows control characters in composition text as escapes
* fix(studio-server): making or serving a still sweeps the transcode cache
Render, capture and the HTML bundler used a narrow private-file rule that
only refused Git data and env files. A composition's script could read the
app's own records under .hyperframes/ (chat history, hand-off, comments,
job log) from the file server, or a data-composition-src could inline one.
Every file server a composition reaches now uses the rule Studio, play and
present already use: Git, env, and .hyperframes/ apart from Studio's own
files and the prepared-assets and requests folders. The rule now matches
Windows spellings too: any case, trailing dots or spaces, backslashes and
stream suffixes.
* feat(studio-server): an opaque preview loads its images, media and fetched files
* refactor(studio-server): the fetch grant's reach is named, not commented
* test(studio-server): the main preview loads under the frame-token path
* refactor(core): Studio's key combo and typing-target rules live in core
* chore(core): raise the packed size budget for the key-combo and typing-target subpaths
* fix(producer): shader transitions blend in Node when the page cannot composite them
* test(producer): the compositor-unavailable fixture uses the usual 30 dB threshold
* fix(engine): restricted downloads use the capture proxy's address rule
With PRODUCER_RESTRICT_PAGE_NETWORK on, remote asset downloads now go
through the same transport and host check as the render page's capture
proxy:
- An IPv6 literal host, or a name the capture proxy refuses, is not
permitted, on the first request and on every redirect hop.
- Names resolve to IPv4 answers only, every answer must be public, and
the download dials the checked address. A name with only IPv6 answers
does not download, as in the capture proxy.
- Proxy variables (HTTPS_PROXY, NO_PROXY) are not used in this mode, so
no host is dialled without the address check. HYPERFRAMES_CAPTURE_PROXY
still routes downloads, as it does for the page.
With the setting off, downloads keep the existing transport.
* fix(engine): restricted downloads keep undici's transport with the capture resolver
The capture transport refuses any compressed response and sends no
User-Agent, and its socket and parse errors were not retried. Restricted
downloads now use the same undici agent as before, with a connect lookup
that calls the capture proxy's resolvePublicAddress: IPv4 answers only,
every answer public, and the checked address is the one dialled.
This keeps the old request headers (User-Agent, Accept: */*), gzip, br
and deflate unpacking for text fetches under the existing byte cap, the
asset path's refusal of compressed bodies, "fetch failed" retry handling,
and connection reuse. Non-public destinations stay non-retryable refusals.
* fix(studio-server): previews with no fetch metadata on a LAN host count as another site
Browsers send no Sec-Fetch-Site header over plain HTTP to a non-loopback
name, such as a LAN address used with HYPERFRAMES_PREVIEW_HOST. The preview
guards now count such a request as the Studio's own page only when its
Referer names the same host. Otherwise the preview is marked and sandboxed,
its validators are dropped, and the thumbnail, waveform, peaks and
frame-token routes refuse it.
* fix(cli): play marks and sandboxes compositions another site loads
The play server's /composition/* route now applies the same another-site
decision as the Studio preview. A load from another site comes back with
the sandbox CSP and without validators, and its HTML carries the
hyperframes-preview-for-another-site mark, so the runtime offers that site
no editing port. The play page's own player is unchanged.
* fix(player): send a same-origin Referer from the composition frame
The player's composition iframe used no-referrer, so on a plain-HTTP LAN
host its own composition load carried neither Sec-Fetch-Site nor Referer
and the Studio and play servers sandboxed it as another site's load.
same-origin sends the host page's URL only to a composition on the same
origin, so another origin still learns nothing about the embedding page.
* fix(cli): benchmark applies the received-files rule for files outside the project
hyperframes benchmark now asks outsideAssetsAllowed() before it creates its
render jobs, the same check hyperframes render and the Studio render route
use. A project whose received-files record lists code files, or whose record
cannot be read, renders without files from outside the project.
* fix(cli): a received .env key never stands in for an expired login
A HeyGen key that a received project's .env set is now used only when no
login or API key is saved. When the saved login has expired, cannot refresh,
belongs to another OAuth client, or HeyGen refused its refresh, the CLI
resolver, afterRefreshFailure and the media-use skill's resolver raise the
login error instead of switching to that key. A person's own .env key and a
shell-exported key keep their current priority.
* fix(cli): a broken credentials file never gives way to a received .env key
A malformed or unreadable ~/.hyperframes/credentials now counts as a saved
credential when deciding whether a received project's .env key may be used.
The CLI resolver raises INVALID_STORE and the media-use skill says the file
can't be read. A person's own .env key keeps its current behaviour. The skill
now tells a person whose saved login belongs to another OAuth client to run
auth login rather than auth refresh.
* fix(cli): cloud render no longer refuses a file only an unused page names
The video's pages are walked from the staged upload, starting at index.html, with the
renderer's resolver; when a page it loads is not staged text, every page stays strict.
* fix(cli): cloud render --composition checks missing files from the page it renders
* fix(core): pages with many CJK characters fetch fonts in several requests
* fix(core): keep text that shapes together in one font request
* fix(core): keep composing characters and their composed forms in one font request
* fix(core): composed forms keep their case variants in the same font request
* fix(cli): uploads leave out outside scripts; .ts counts as video only when it is one
* chore(studio-server): raise unpacked size budget for the upload file-type rule
* fix(studio-server): transport stream check never waits on a pipe and needs binary bytes
* feat(cli): uploads list every file they carry from outside the project
* perf(lint): scan CSS url references in linear time; the test checks growth, not wall time
* test(lint): cap one large malformed stylesheet instead of comparing two small timings
A film clip whose src resolves to a UNC or device path outside the
project's own share is now reported as "network share paths are not
analyzed". The clip is never stat'ed, probed, or decoded. Relative and
local absolute srcs resolve as before.
* fix(cli): uploads leave out outside scripts; .ts counts as video only when it is one
* chore(studio-server): raise unpacked size budget for the upload file-type rule
* fix(studio-server): transport stream check never waits on a pipe and needs binary bytes
* fix(cli): a subfolder of an imported project keeps its received-files record
Publish, cloud render and feedback now read the received-files record at
the nearest folder at or above the one they stage, so staging a subfolder
of an imported project still leaves out the outside files its received
files reference.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): a received file counts under any spelling of the same file on disk
The received-files lookup now also matches a project file that is the
same file on disk as a listed path, so a name respelled in letter case on
a case-insensitive disk keeps its received status.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): cloud render prepares only the GIFs staging kept
The entry GIF prep before a cloud render upload now reads only the staged
copies it is handed. An image whose source staging left out keeps its
reference as written and is never read from disk.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli,studio-server): never hand a playlist named as media to ffmpeg
A media file whose first bytes are an HLS playlist, an XML manifest or a
concat script now stays out of ffprobe and ffmpeg: cloud render leaves it
in the zip as authored, the media probe reports it as not media, and the
proxy copy refuses it. Real media never starts with these bytes, so it
probes and transcodes as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(studio-server): catch playlists with control bytes, and keep them from stills and audio probes
A text playlist or concat script is detected whatever bytes follow its
first line; only markup must be all text. A still frame and the audio
probe behind waveforms, peaks and loudness now refuse a file the probe
reads as not media, as the proxy copy already does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): a parent folder's state the CLI cannot read never marks a project received
The walk up for a received-files record now skips a parent whose record
cannot be reached. Only a record file that is there but cannot be read
or parsed counts every file as received.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): a drive that reports no file ids matches no received file by identity
Some network and virtual drives report a file id of 0 for every file, so
the same-file match now gives such a file no identity instead of
matching it to every listed file.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor(studio-server): read a media file's first bytes with a read stream
The playlist check reads the first 64 bytes through a bounded read
stream instead of a raw file descriptor, and the audio-track probe's doc
comment sits on its function again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* chore(studio-server): raise the packed size budget for the media playlist check
Main packs @hyperframes/studio-server close to the 1% tolerance on the
168500 budget, and the playlist check adds about 320 bytes, which put
the Build over. Set the budget to 172000.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli,studio-server): open media only through self-contained media demuxers
Every ffprobe and ffmpeg input that cloud render offload, the publish
proxy bake and Studio's media helpers open now carries a demuxer
whitelist built from the supported media extensions: MP4/MOV/M4A/AVIF,
Matroska/WebM, MXF, AVI, MPEG-PS/TS, FLV, ASF, Ogg, WAV, MP3, AAC, FLAC,
AIFF and the still-image formats. Playlist, manifest and concat demuxers
can no longer open, whatever a file's bytes say. The proxy's own concat
step is left as it is.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli,studio-server): keep media probe argv flat, and expect the demuxer list in loudness args
Two probe argv literals built from nested spreads are flat again, so
"--" stays the visible second-to-last entry the argv contract checks.
The normalize-audio test expects the demuxer list before the input.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): play and present answer only local host names and allowed hosts
play and present listen on every interface. Each request now passes the
same Host check preview uses: localhost, loopback addresses and this
machine's own addresses and names are answered, and any other host name
gets a 403 naming HYPERFRAMES_PREVIEW_ALLOWED_HOSTS until it is listed
there. Phones and other machines on the network still open them by LAN
address or .local name.
* fix(cli): a local render of a received composition leaves out files outside the project
When the composition a local render opens is listed in the project's
received-files record, hyperframes render (full and range) no longer
brings in files outside the project. It prints one line naming the
references it leaves out and why. An unreadable record counts as
received. A project whose entry is the person's own, or that has no
record, renders exactly as before.
* fix(cli): received-render scan reads safely and treats an unreadable record as unknown
The scan behind the received-composition notice skips a file it cannot
read, finds url() references with a linear forward scan, and reads at
most 2 MB of each file. An unreadable received-files record no longer
counts as everything received for local render: the render keeps
outside files and names the record it could not read. Publish and cloud
upload still treat it as received. The notice now says the render stops
on a sub-composition outside the project rather than leaving it out.
* fix(cli): the studio render of a received composition leaves out files outside the project
The Studio render route applies the same rule as hyperframes render:
when the composition it renders is listed in the received-files record,
the job does not bring in files outside the project, and the server log
names the references it refuses.
* chore(cli): name the unreadable received-files record helper without a comment
* fix(cli): local render refuses outside files when any composition it plays was received
The received-file rule for local render (CLI and Studio) now checks
every page and stylesheet the render pulls in, not only the entry, so
an own index.html that plays a received scene also reads nothing
outside the project, and the notice names the received file. An
unreadable, truncated or empty received-files record refuses outside
files too, as publish and cloud upload do, and the notice names the
record it could not read.
* fix(cli): decide a received render from the files the compiler reads
Local render (CLI and Studio) now leaves outside-file handling as it is
for a project with no received-files record or an empty one, without
scanning anything. When the record lists files, the compiler's own read
list for the composition decides: the bundler runs out of process with a
deadline and reports every project file it reads, so stylesheet imports
and files past any scan limit count. A received file in that list, a
build that fails or runs past its deadline, or an unreadable record
refuses outside files. The scan that names refused references in the
notice is advisory only: it guards every file system call, tracks visited
files by real path, and caps the files and bytes it reads.
* fix(cli,producer): decide a received render with the render compiler's own composition traversal
The producer exports renderCompositionFiles, which walks
data-composition-src with the same resolution, normalization and private
file rules the render's sub-composition pass uses; that pass now calls
the same helper. The received-render decision adds those files to the
bundler's read list, so leading-slash and percent-encoded references the
render reads count. The advisory scan resolves compositions the same way
and reads template contents.
* fix(cli): refuse outside files in a local render whenever the project has received files
A local render (CLI and Studio) reads nothing outside the project when
the project's received-files record lists any file, or cannot be read.
No static list matches what a render consumes: the page loads its
stylesheets and scripts through the browser at capture time, and
scripts can load more, so the decision no longer depends on one. The
reference scan stays only to name what the notice says is left out. This
drops the out-of-process bundler read list and the producer composition
walk added for that decision.
* fix(cli): refuse outside files only when a received file can load other files
A local render (CLI and Studio) reads nothing outside the project when
the received-files record cannot be read, or lists a received file that
can reference others: an .html, .htm, .css, .js, .mjs, .cjs or .svg
name, or content that reads as HTML, SVG or CSS. A project that only
received images, video, audio or fonts keeps bringing in outside files
as before.
* fix(cli): keep outside files only when every received file is positively media, a font or an image
A local render keeps bringing in outside files only when each file the
received-files record lists has a known media, font or image extension
and starts with that type's signature bytes. Any other received file, a
listed file that is missing or unreadable, or a mismatch between name
and content refuses outside files. An ancestor record's entries now
match a subfolder whose name is spelled in another Unicode form; the
entry keeps its own spelling on disk.
* fix(producer,core): a sub-composition must be an .html file
The render compiler and the runtime load a data-composition-src only
when it names an .html or .htm file. Any other file, such as an image or
video, is refused as a page: render stops naming it, and the runtime
reports the load failure without fetching it.
* fix(producer,cli): the render and static file servers forbid type sniffing
Both servers send X-Content-Type-Options: nosniff with every response,
so the browser loads a file only as the type its extension names: an
image or video cannot run as a script or apply as a stylesheet. .cjs is
served as JavaScript.
* test(cli): a received media file in a composition slot is refused as a page
* test(cli): an empty nested received-files record cannot hide an outer record's received page
* fix(cli): a received media file reached under another name refuses outside files
The received-render decision now compares file identities, not names.
It walks the project once, bounded by entry count and time, and refuses
outside files when a hardlink or a symlink with a name that is not
media reaches a received file, or when the walk cannot finish. A
received file with no identity matches nothing, as in the received
files check.
* fix(cli): compare received files by link target and keep outside files for an empty record
A received media file and every other project name are now compared by the
identity of the file they reach with links followed. A linked folder that
leads outside the project, or a file whose identity cannot be read, counts as
unchecked and refuses outside files. An empty received-files record keeps
outside files without walking the project.
* chore: raise the producer unpacked size budget for the composition file checks
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): add lists files from a received config's registry as received
When a project's hyperframes.json is listed in .hyperframes/received-files.json
and names a registry other than the public one, add now lists every file it is
about to write in that record before writing any of them. This covers every
caller of addToProject: add by name, by tag, several names, and Studio.
The decision reads the config the install uses, so a byte-order mark or another
spelling of the file on disk is handled the same way. A person's own config, the
public registry and a record that cannot be read are left as they are.
* fix(cli): list received files as they are written, under a shared lock
add now lists each file in the received-files record just before it writes
it, using the write-time decision, so a file kept because the person edited it
during the download is never listed. When an install fails, it takes back the
entries it listed but never wrote.
Changes to the record take .hyperframes/received-files.json.lock, created
exclusively. A writer retries for about 2 s, takes over a lock older than
about 10 s, and always releases it. A writer that cannot take the lock in time
fails with a clear error instead of writing past it.
* fix(cli): an unreadable config counts as received; a failed install keeps what is on disk
add now counts an install as received when hyperframes.json exists but cannot be
read or parsed, since that config is not proven to be the person's own. When an
install fails, it takes back only the entries it listed whose files are not on
disk; files that landed stay listed.
* fix(cli): the received-files lock resolves the project folder as the installer does
The record's lock and write resolved the project folder with fs.realpathSync,
while the containment check resolves targets with the native realpath. The two
can spell the same folder differently: a Windows short name or a folder typed
in another letter case. The target then looked outside the project and the
change was refused. The record now resolves the folder with the installer's
realProjectRoot, so the root and target compare in one spelling.
The failed-install tests now fail the write by stubbing it, which works on every
platform, instead of a read-only folder, which Windows ignores.
* fix(cli): batch render outputs stay inside the --output folder
A row value may still name a subfolder of the folder written before the
template's first placeholder, but a value that resolves outside it stops
the batch before any folder is made or file is written.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): project text shows control characters as visible escapes
compositions, lint, snapshot and info --media print composition ids, sources,
element ids, lint messages and media paths through one helper that writes
C0, DEL and C1 control characters out as \xNN, so they reach the terminal as
text. The CLI's own colors are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(studio-server): waveform and peaks key one cache entry per file and refuse other sites
Both caches are keyed on the resolved project-relative path, so another
spelling of the same file (a doubled or trailing slash) reads the cached
peaks instead of decoding again. Waveform and peaks requests that another
site makes are refused the way thumbnail requests already are; the Studio
page asks for both as itself.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(studio-server): preview stills run at most four at a time
A video's stand-in still is made through a small job queue: four at once,
32 more waiting, and a 503 with Retry-After beyond that. A waiting request
whose client goes away leaves the queue without starting ffmpeg.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): batch output folder check keeps backslash names and row-named paths
A trailing backslash counts as a separator only where it is one, and the
check compares path segments. A template that starts with a placeholder names
no folder, so each row's value is the whole output path as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): lint header shows control characters in file names as escapes
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(studio-server): a finished or in-progress still never waits for a still slot
Only a still that is neither on disk nor already being made takes one of
the four slots, through a hook resolveStill gives the route. Waiters queue
without a cap and nothing answers 503, since a poster is never asked for
again. A waiter whose client leaves still never starts ffmpeg; a request
that shared that waiter's still tries again on its own.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* chore: raise the studio-server packed budget to 172000
main already packs about 170 KB against a 168500 budget; the request guards,
still slots and cache-key change here add about 550 bytes.
* fix(core,studio-server): a preview served to another site offers no editing port
When the Studio server answers a preview document to another site's
request, it now marks the document, and the runtime's editing agent
ignores every connect in a marked document. The Studio's own frames,
which load the preview from the same origin, keep the editing port.
The marked copy carries no ETag, so it is never revalidated in place of
the unmarked one.
* fix(studio-server): a media still stops when no request waits on it
Each still job now counts the requests waiting on it. A request whose
connection closes leaves the job, and when the last one leaves, the
job's ffmpeg is killed and the job is forgotten, so the next request
starts afresh. A caller that passes no signal keeps the job running,
and a request that has already closed starts nothing.
* test(studio-server): group the still cancellation tests after the real-ffmpeg still test
* fix(studio-server): a preview framed by another site is never revalidated
A frame load of a preview from another site now drops If-None-Match and
If-Modified-Since before the route answers, so it always gets a fresh,
marked document instead of a 304 that could reuse the unmarked copy a
browser cached for the Studio's own page on the same site. Subresource
loads keep their validators.
* fix(cli): check findings and keyframes headers show control characters as escapes
check prints each finding's message, source file, selector and fix hint, and
keyframes prints each composition's name and source, through terminalText.
* test(studio-server): a request sharing an abandoned still waiter gets its still
* chore(studio-server): raise the unpacked size budget for the still job slots and cross-site guards
* fix(cli): layout and contrast lines in check show control characters as escapes
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(studio-server): replace lock start records by rename
The owner lock's start records (owner.pid.start, owner.pid.evict.start) and
the daily prune stamp are written through replaceFileAtomically: an exclusive
temp sibling renamed over the name. A symlink, hard link or other entry already
at that name is replaced rather than written through.
Lock files (owner.pid, its start record, owner-url.json) are read with
O_NOFOLLOW and O_NONBLOCK and only when they are regular files, so a link is
not followed and a FIFO is not waited on.
* fix(studio-server): clear lock records before replacing them
The start records and the daily prune stamp are removed before the rename,
so a symlink to a read-only file or a read-only file at that name no longer
fails the write. The prune stamp is read without following links or waiting
on a FIFO, like the other lock files.
* perf(parsers): find CSS url() references in linear time
A new css-url-scans entry finds what each CSS url( regex in the compile path
and the runtime finds, match for match and group for group, with forward
scans. The regexes restart from every unclosed url(, @import or background:
and read to the end of the text each time, so their time grows with the
square or cube of the input. The scans read each character a bounded number
of times.
rewriteCssAssetUrls uses the scan in place of CSS_URL_RE, which stays
exported for existing callers. A fuzz test checks 20,000 generated pages
against the regexes, and timed tests cover 1 MB of each unclosed form.
The parsers packed-size budget rises to cover the new entry.
* perf(core): bundler, SVG id namespacing and runtime find CSS urls in linear time
The bundler's url() inlining, CSS-file rebasing and @import inlining and
hoisting, the SVG url(#id) rewrite, the runtime's sub-composition url()
rewrite and the scene swap's url and @import list now use the
@hyperframes/parsers css-url-scans entry. Each returns the same matches as
the regex it replaces, in time linear in the CSS length.
* perf(producer): compile finds CSS urls, imports and backgrounds in linear time
Asset collection, head-style @import coalescing, the ancestor
background-image check, remote background and font-face url localization and
local font embedding now use the @hyperframes/parsers css-url-scans entry,
which returns the same matches as the regexes it replaces. A timed test
covers remote background discovery on 200 KB of unclosed background urls.
* perf(parsers): scan style blocks, font-face blocks, comments and mask urls in linear time
css-url-scans gains forward scans for `<style>` blocks, `@font-face { }`
blocks, `/* */` comments, the quoted-or-bare `url()` form and the
`mask-image: … url()` declaration. Each returns the same matches and groups
as the regex in its doc comment. The fuzz test now checks all fifteen scans
on 20,000 generated pages, and the timed tests cover 1 MB of each unclosed
form.
* perf(core): strip and restore stylesheet comments in linear time
Bundling a stylesheet finds its comments with the parsers comment scan and
puts them back in one pass over the placeholders. A comment comes back
verbatim, including text such as `$'` that a string replacement pattern
would expand.
* perf(producer): find style and font-face blocks in linear time
Remote font discovery and local font embedding read `<style>` blocks and
`@font-face` rules with the parsers scans, which return the same blocks as
the regexes they replace.
* perf(lint): read stylesheet and mask-image urls in linear time
The missing stylesheet asset and texture mask rules find `url()` values with
the parsers scans, including the mask-image fallback for a stylesheet that
does not parse.
* perf(cli): rewrite published CSS urls in linear time
Publishing rewrites stylesheet `url()` references with the parsers scan in
place of CSS_URL_RE, with the same matches.
* chore: size budgets cover the CSS scans in parsers and inspector
verify:packed-manifests measures parsers at 894,913 unpacked bytes and
inspector at 3,764,456, both just over 1% above their budgets once the
linear CSS scans are bundled. The budgets rise to 900,000 and 3,790,000.
* perf(parsers): scan stylesheet links and CSS rules in linear time
css-url-scans gains forward scans for `<link … rel="stylesheet" …>` tags and
`selector { declarations }` rules that return the same matches and groups as
the regexes in their doc comments. Both join the 20,000-page fuzz test and
the 1 MB timed tests.
* perf(core): restore stylesheet imports and list scene font faces in linear time
Bundling a stylesheet puts its comment and @import placeholders back in one
pass, so an imported sheet comes back verbatim and is not searched again.
fontFaceBlocks lists @font-face rules with the parsers scan. A fuzz test
checks the one-pass restore against the sequential one on 20,000 sheets.
* perf(producer): find stylesheet links, CSS rules and font-face blocks in linear time
Font stylesheet inlining and project font discovery find stylesheet links,
and the ancestor background check finds CSS rules, with the parsers scans.
extractFontFaceBlocks matches every brace in one pass instead of counting
depth again from each @font-face; a fuzz test checks it against the depth
count on 20,000 sheets.
* test(core): time the CSS placeholder restore directly
The bundle test with 6,000 imports was timed mostly on 6,000 file reads,
which took 2.7 s on the Windows runner. The restore is now timed on its own
over 20,000 placeholders, and a short bundle test still checks that every
import of a sheet is inlined.
* fix(producer): keep the render server local unless a token guards it
startServer listened on every interface and trusted any caller without
Origin/Sec-Fetch-Site headers, while /render, /render/stream and /lint used
caller-supplied projectDir, entryFile and outputPath as arbitrary paths.
- Listen on 127.0.0.1 by default (hostname / PRODUCER_HOST). A non-loopback
address needs authToken / PRODUCER_AUTH_TOKEN, which createProducerApp then
requires as a bearer token on every route but /health. Log the real address.
- Confine request paths, symlinks included: projectDir to projectsDir
(PRODUCER_PROJECTS_DIR, default cwd), entryFile to the project, outputPath
to rendersDir, checked before any directory is created. The debug dir is
never project content.
- Refuse debug: true unless allowDebug / PRODUCER_ALLOW_DEBUG=1, drop
perf.renderId from responses, and stop echoing resolved absolute paths.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LqRB2bN8KsmaW2KUs9k6cb
* fix(producer): apply the request policy at the public app, not the shared handlers
createRenderHandlers is also mounted by trusted embedded services that pass
absolute temp project and output paths and no projectsDir. Confining it by
default turned their renders into 400s.
- New confineRequests option: createProducerApp and startServer default it to
true; createRenderHandlers keeps its old contract unless a caller opts in.
- Under the policy, /lint refuses an entry that leaves the project through a
symlink or reaches the debug dir, through an optional guard on
prepareHyperframeLintBody (trusted callers are unchanged).
- A blank auth token counts as no token, so it no longer unlocks a
non-loopback listen address.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LqRB2bN8KsmaW2KUs9k6cb
* chore(producer): trim added comments and raise the producer size budget
The comment ratchet refused the comment lines the hardening added to
server.ts and hyperframeLint.ts; they are cut or folded into one line.
The producer tarball grew to 900392 bytes packed (budget 890000, 1%
tolerance) with the auth middleware, request policy and lint entry guard,
which ship in each of the producer's server bundles. The budget moves to
901000 to cover that code.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LqRB2bN8KsmaW2KUs9k6cb
---------
Co-authored-by: Claude <noreply@anthropic.com>
* fix(engine): the media playlist check reads HLS lines and attributes as ffmpeg does
The folder check now finds every name ffmpeg's hls demuxer would open:
- lines end at CR, LF or NUL, and a line is also read cut at ffmpeg's
4095-byte limit;
- URI attributes are split like ff_parse_key_value: quoted or bare values,
with backslash escapes removed;
- leading ID3v2 tags are skipped before the playlist signature, and a file
behind more than 16 of them is refused.
Each refused shape is checked against ffprobe 8 in the test: ffprobe opens
the outside file, and playlistLeavesItsFolder now refuses the list.
* fix(cli): snapshot never hands FFmpeg a project video that is a document or an escaping playlist
`hyperframes snapshot` decodes held project videos with FFmpeg and FFprobe.
It now runs the engine's media check first: a file that is a document, or an
HLS or ffconcat list naming files outside its folder (under any extension),
is skipped with a "not media" warning and never reaches FFprobe or FFmpeg.
Real clips and HLS lists of segments beside them snapshot as before.
* fix(engine): playlist check reads lines as bytes and refuses unreadable nested lists
- Playlist text is read as latin1, one character per byte, so a line is cut
at 4095 bytes exactly where ffmpeg cuts it, even when it holds bytes that
are not UTF-8. Nested names are opened by those same bytes.
- A nested name that exists but cannot be read is refused; a name that does
not exist, or is too long to exist, is still kept.
- Each name is checked once, each file is opened once (ID3 header and head
in one read), and each file's verdict is reused within a call: 100k
segments naming one file drop from seconds to tens of milliseconds.
* perf(cli): snapshot checks each project video once and warns once when it is refused
* test(engine): the ffprobe playlist proof ends its options with -- before the input
* chore(engine): raise the packed size budget to 500000 bytes
The byte-exact HLS playlist parser adds about 6 KB to the packed engine
(496647 in CI against a 490452 budget).
* fix(engine): media inspection never waits on a FIFO or device
The document sniff and the playlist walk open with O_RDONLY|O_NONBLOCK where
the platform has it, fstat the descriptor, and refuse anything that is not a
regular file before reading; a folder still fails as it did. A FIFO with no
writer, as a media source or named in a list, is refused at once.
* fix(cli): snapshot inspects project videos within the probe deadline and its cleanup scope
The held-video hints are gathered inside the capture's cleanup scope, and
every media inspection gives up after the same 30 s limit as the FFprobe
call it guards, treating the file as not media.
* fix(engine): read PRODUCER_RESTRICT_PAGE_NETWORK's usual spellings and restrict on an unknown value
true, 1, yes and on turn the render network restriction on and false, 0,
no and off leave it off, in any letter case and with surrounding
whitespace. Unset or blank stays off for local renders. Any other value
now restricts and logs a warning once, instead of leaving the render
browser and ffmpeg URL inputs unrestricted.
* fix(producer): on a render service the audio probe reads a composition URL from a downloaded copy
With PRODUCER_RESTRICT_PAGE_NETWORK on, a video URL the composition
names is fetched through the shared downloader (resolved-address check,
pinned connection, per-hop redirect checks) before ffprobe reads it, and
skipped if the download is refused. In a distributed audio step the copy
lands in the mixer's work folder, so the mix reuses it instead of
downloading again. Without the flag the probe reads the URL as before.
* fix(engine): on a render service a URL that ffmpeg reads in place must be a self-contained container
With PRODUCER_RESTRICT_PAGE_NETWORK on, every https input ffmpeg or
ffprobe opens directly (such as a signed storage URL for a project's
video or audio sidecar) also gets -format_whitelist with the same
self-contained containers the Studio media helpers allow. A playlist,
manifest or concat list stored under a media file name can no longer
make ffmpeg open other network locations. Local renders are unchanged.
* fix(producer): the render-service audio probe skips only refused URLs and downloads once
A composition URL is skipped, with one warning, only when the download
address rules refuse it. Timeouts, HTTP errors and a cancelled render
fail the probe as before. Videos without sound, muted or hidden are not
downloaded. The in-process extract stage passes the render's abort
signal and the mixer's work folder, so the mix reuses the copy.
* fix(producer): unit tests load the orchestrator again; raise the engine size budget
audioWorkDir moves to render/shared.ts, so tests that replace the audio
stage module still load the orchestrator. The format-whitelist test
probe ends its options with -- as the ffprobe argv contract requires.
The engine packed size budget rises to 500000 bytes.
* fix(registry): catalog payload check passes a missing caption-overrides.json
The runtime requests an optional caption-overrides.json on any page with a
.caption-group and reads a 404 as "no overrides". The docs host has no such
file, so every caption payload that uses .caption-group (caption-emoji-pop,
caption-neon-accent, caption-pill-karaoke, caption-weight-shift) failed
verify:catalog-payloads with a 404, its console line, and an aborted request.
The check only ran on those payloads once a change regenerated them.
The verifier now sets that miss aside, as `hyperframes validate` already does
through shouldIgnoreHttpError. A non-404 status, any other failure on that URL,
and an invalid overrides file still fail. To match the console line by URL,
checkPageLoads now appends the resource URL to Chrome's "Failed to load
resource" message, whose text leaves it out.
* style(producer): name the failed resource URL instead of commenting it
* fix(cli): printed Windows commands read folder names literally in PowerShell, cmd and Git Bash
shellArg's Windows branch wrapped any non-plain name in double quotes,
where PowerShell and Git Bash still expand $ and backticks. Names with
those characters (or PowerShell's curly quotes) are now single-quoted,
which both shells read literally; a name that cmd.exe would also split
on & | < > ^ gets no printed path, and the hint says to run the command
from the project folder. A bare comma or leading @ is quoted so
PowerShell does not read an array or a splat. Ordinary names with
spaces, accents and apostrophes print as before.
The background preview's stop footer uses the same helper instead of
JSON quoting, which left $(...) live in POSIX shells and PowerShell.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): printed Windows paths with backslashes keep them in Git Bash
A Windows word containing a backslash printed bare, and Git Bash strips
unquoted backslashes, so C:\Users\Bob\film reached the CLI as
C:UsersBobfilm. Such words now take double quotes, which PowerShell,
cmd.exe and Git Bash all read as written; trailing backslashes are
doubled so they do not escape the closing quote.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): open --json gives a null stopCommand when the folder has no safe spelling
When no spelling of the project folder is safe to paste in every Windows
shell, the left-running preview's stopCommand held prose. It is now null,
and stopCommandFromProject names the command to run from inside the
project folder. The printed line is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* chore(cli): drop a comment the nullable return type already states
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): a Windows name with no spelling common to every shell prints no path
PowerShell single quotes do not group a word in cmd.exe, and Git Bash
reads 'it''s' as its, so the single-quoted form printed for names with
$, a backtick or a curly double quote was wrong in two of the three
shells. Those names now take the existing null fallback, and the hint
says to run the command from the project folder.
Round-trip tests check that every printed Windows word reaches the
program unchanged: through Git Bash everywhere, and through cmd.exe and
PowerShell on Windows, where the CLI's Windows CI job now runs them.
The names cover apostrophes, typographic quotes, mixed quotes, cmd
operators and trailing backslashes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(cli): run the cmd.exe round-trip through spawnSync
execFileSync's options type has no windowsVerbatimArguments, so the
Windows-only cmd.exe test failed typecheck. spawnSync takes the option,
and the test now also checks the exit status.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(cli): expect the quoted Windows path in the relative open command
On Windows a relative path holds a backslash, which shellArg now prints in
double quotes so Git Bash keeps it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): a Windows name with %NAME% or ! prints no path
cmd.exe expands %NAME% inside double quotes, and interactive Git Bash
history-expands ! there, so neither name reads literally in every
Windows shell. Both now take the null fallback. A lone % such as 100%
still prints.
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(cli): a file is received when any record at or above the staged folder lists it
A received-files record inside a staged subfolder could list nothing and
hide the project's own record, so outside references from received files
were copied into the upload again. Every record from the staged folder up
is now read: a nested record can add received files but never remove them.
* fix(cli): a HeyGen key from a project .env never replaces a saved credential
A HEYGEN_API_KEY or HYPERFRAMES_API_KEY that only the folder's .env set now
ranks after the person's stored login and saved API key, in the CLI resolver
(cloud render, publish, media helpers) and in the media-use skill scripts.
With no saved credential the project key still works, and the first use
prints one line naming the variable and the .env it came from. A key
exported in the shell keeps its current precedence.
* fix(cli): capture's agent files count the page's fonts instead of naming them
Font family names come from the captured page, and a short phrase passed the
plain-name check into AGENTS.md, CLAUDE.md and .cursorrules. The Brand Summary
now gives the number of font families and points to extracted/tokens.json,
which already holds the names as data.
* fix(cli): only a received project .env key gives way to a saved credential
A HeyGen key from a project .env now comes after the stored login and
saved API key only when that .env came in with a received project: a
received-files record at or above the folder lists it, or a record there
cannot be read. A person's own .env keeps its priority. The media-use skill
scripts read the same records and apply the same rule.
* fix(cli): a shell key wins over a received .env key under either name
* docs(cli): a received project's .env key comes after the saved login and key
* fix(cli): drop the duplicate font count from capture's agent files
The Brand Summary repeated the font count the file table already gives,
and counted entries that are not font names. The table row stays.
* fix(cli): name the received-files record in the skill comment without a path
* fix(gcp-cloud-run): render pages cannot reach the service's own dispatch endpoint
The Cloud Run image now sets PRODUCER_RESTRICT_PAGE_NETWORK=true, so a
rendered page reaches only its own project server and public addresses,
not loopback ports or the metadata server. The dispatch endpoint also
refuses any POST that carries an Origin header: browsers send one on every
POST and Cloud Workflows never does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(engine): the default frame cache folder is private to the user who renders
The default extraction cache sits in the shared temp folder. It is now
created with mode 0700, an existing one owned by the same user is closed
to others (its entries stay, so cache hits carry over), and one that is a
link or belongs to another user is not used: that render runs without the
cache and prints a warning. The animated GIF cache under the same folder
follows the same rule. A folder set with HYPERFRAMES_EXTRACT_CACHE_DIR is
left as it is.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(gcp-cloud-run): a Terraform variable turns the page network restriction off
restrict_page_network (default true) sets PRODUCER_RESTRICT_PAGE_NETWORK on
the service. The package README and docs say what the restriction refuses
and how to turn it off for trusted compositions that need private hosts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(engine): restore an unset TMPDIR as unset after the cache-root tests
Assigning undefined to process.env stores the string "undefined", so on a
runner with no TMPDIR every later test made its temp folder under ./undefined.
* test(engine): the frame-extractor link test also restores an unset TMPDIR as unset
* chore(engine): raise the packed size budget to 500000
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
caption-emoji-pop and caption-neon-accent look a word up in KEYWORDS and
WORD_EMOJI after stripping everything outside a-z, and caption-glitch-rgb
counts a word's \w characters to decide on the aftershock. A Korean, Cyrillic
or accented word comes out empty or cut, so a keyword written in that script
never matches and a long word never pulses.
The three components now strip by Unicode letter (\p{L}) instead, which is
the same set for a lowercased English transcript.
Co-authored-by: Claude Fable 5.2 <noreply@anthropic.com>
* fix(deps): refresh zod, smol-toml, aws-cdk-lib and related resolutions in bun.lock
Lock-only re-resolution within existing ranges:
- zod 4.4.3 -> 4.6.7 (GHSA-mrhx-c7xc-m3w5)
- smol-toml 1.6.1 -> 1.9.0 (GHSA-7w5x-hrqm-74c2, GHSA-r4xh-jqrq-34v2)
- aws-cdk-lib 2.254.0 -> 2.260.0 (GHSA-vcrf-j523-4mrf)
- fast-xml-parser 5.7.3 -> 5.12.0 (GHSA-474x-hrwr-fgx4)
- @babel/core 7.29.0 -> 7.29.6 (GHSA-4x5r-pxfx-6jf8)
- postcss-selector-parser 7.1.7 -> 7.1.8
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* chore(inspector): raise unpacked size budget for postcss-selector-parser 7.1.8
The inspector bundles postcss-selector-parser; 7.1.8 inlines its escaping
(drops cssesc) and adds the hardened parser paths, which grows the unpacked
inspector bundle by about 10 KB and tips it past the 1% tolerance.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(deps): bump @mcp-b/global to 6.1.0 to clear @modelcontextprotocol/server advisories
@mcp-b/global 5.x and 6.0.x pin @modelcontextprotocol/server 2.0.0, which is
affected by GHSA-f897-62hf-6gxq, GHSA-46jr-4wx8-v453, GHSA-jm63-cq47-j5jp and
GHSA-rvq5-wwqv-78pq (>=2.0.0 <2.3.0). 6.1.0 is the first release that pins
server 2.4.0.
6.x installs the W3C WebMCP polyfill and passes { signal } to execute; the
real-package tests now drive the 6.x surface (secure context, MCP tools/call).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(studio): close the WebMCP polyfill after the real-package suite
6.x's BrowserMcpServer sends a tools-changed event on a timer after the
last tool unregisters. Without cleanupWebModelContext() it fired after jsdom
teardown and vitest failed the run on an unhandled rejection. Also scope the
transport test's header to the MCP transports it actually checks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
When one capture worker fails, the pool aborts its peers and each of them
reports "Parallel worker cancelled". The pool message listed every worker in
id order, so the one real cause sat between cancellations:
Worker 0: Parallel worker cancelled; Worker 1: Parallel worker cancelled;
Worker 2: Disk capture may need ~2712.6 MB ...; Worker 3: Parallel worker cancelled
The message now carries the failures that are not cancellations and counts
the peers that were stopped because of them. A pool where every worker was
cancelled keeps the full list.
Co-authored-by: Claude Fable 5.2 <noreply@anthropic.com>
* fix(engine): render refusals name the address an IPv6-mapped or NAT64 host carries
* chore(engine): raise the engine size budget for bounded downloads and refusal classes