* perf(cli): give render the heap its auto workers need in small containers
Auto worker sizing caps capture workers at (V8 heap - 1024 MB) / 640 MB.
Node sizes its default heap from visible memory, so in a 10 GB container
the heap is 2,096 MB and every render captures with one worker, whatever
the core count.
When no --workers is given, no --docker, no heap was set by hand, and the
heap is the only bound holding workers back, render relaunches itself once
with --max-old-space-size set to what CPU and memory allow (capped at half
of memory). The engine exposes heapMbForAutoWorkers() for this.
* fix(cli): count a heap-relaunched render once, keep detached renders alive
The relaunched child skips the CLI's command telemetry, update and skills
notices, and app-history notice; the parent already records and prints
them. A detached render no longer gets SIGHUP forwarded, which its
default handler would treat as a kill. A heap set as --max_old_space_size
counts as set by hand.
* fix(engine): keep the desktop app's per-worker heap figure at 640 MB
Inside Electron the heap is a hard 4,192 MB cage, and that is where the
field OOM behind the old figure happened (#5149: 5 auto workers on a heavy
export). The new 320 MB figure is measured only in a system Node, so the
desktop app keeps 640 MB (4 workers there) until a heavy desktop export is
measured. Containers and plain Node keep 320.
* fix(engine): know the packaged desktop app by its client tag, which runs the CLI on plain Node
* perf(engine): fast capture stays on for frames a shadow, blur or clip effect does not touch
A single drop-shadow, blur, backdrop-filter or blend element, or one clip-path or mask
tween, sent the whole render to screenshot capture, 2-3x slower than drawElement. The two
init gates now name the frames the effect can show on. Only those frames capture by
screenshot, in a second browser opened on the first of them; every other frame keeps
drawElement.
* fix(engine): scan effects in a throwaway session; hand risk frames over only where captured
The effect scan seeked the render's own page to the end, leaving its timeline state ahead of
the render. It now runs in a throwaway session with the render's own seek. A hidden effect
element with a visible descendant counts as shown. Per-frame capture is opt-in per caller
(deFrameFallback), so disk, segmented, chunk and parallel-worker sessions keep the
whole-render fallback, and a route that cannot hand frames over drops drawElement up front.
* fix(engine): a timeline callback keeps the whole-render fallback; trim comment share
A tl.call or onComplete can add an effect's class to any element mid-render, so the elements
found at the start are not the only ones it can reach. Such a timeline is unbounded and keeps
the whole-render screenshot fallback.
* style(engine): shorten the live-frame scan comment
* perf(engine): after a timeline callback, not before, every frame may show the effect
A callback can add an effect's class to any element, so only the frames from the first
callback on lose their element bounds. Frames before it keep the scanned bounds, so a late
callback no longer sends the whole render to screenshot capture.
* perf(engine): open the second screenshot browser only when memory is available for it
Per-frame capture needs a second browser (479 MB measured at 1080p). When the render would
leave less than 1.5 GB available after it, the render keeps the whole-render screenshot
fallback instead. getAvailableMemoryMb moves from the CLI to the engine's system-memory
module, which already owns the host's memory readings; os.freemem() reads 367 MB on this
49 GB Mac, so the guard uses free, inactive, speculative and purgeable pages.
* fix(engine): budget the whole 1 GB per-frame capture adds, not just the second browser
The render's process tree goes from 1002 to 2001 MB at 1080p: the screenshot browser (479 MB)
plus a drawElement browser that is itself about 500 MB larger than a screenshot one. The guard
budgeted 500 MB. It now budgets the 1 GB measured, scaled by output pixels.
* fix(studio): text in Studio reads without em or en dashes
* fix(studio): the dash check also catches HTML entities, and peak reads Unavailable
* test(studio): the dash check test covers every entity form
* fix(studio): messages Studio shows from lint, sdk, core and the renderer drop dashes too
* fix(producer): the gate check stops early on the engine's new fallback wording
* fix: bound HDR and video extraction resources
* fix: trim negative video extraction preroll
* fix: skip invisible video extraction windows
* fix: preserve negative-start loop and held tails
* fix: cap finite video slots to source duration
* fix: bound held-tail frame extraction
* fix: plan from playable video duration
* fix: preserve open-ended held video tails
* fix: resolve held tails from decoded frames
* fix: normalize final-frame probe timestamps
* fix: handle unseekable final-frame sources
* fix: dedupe final-frame probes per render
* refactor: clarify output dynamic range contract
getSystemTotalMb returned os.totalmem() — the host's physical RAM — so a
4GB Docker container on a 32GB host never auto-flagged as low-memory and
the low-memory render profile didn't activate exactly where it's needed
most. Read the cgroup v2 limit (/sys/fs/cgroup/memory.max, with the v1
fallback and its no-limit sentinel handled) and use min(host, cgroup).
The probe is best-effort and non-Linux platforms never touch /sys.
Review follow-ups: worker sizing (calculateOptimalWorkers) and the
getSystemResources diagnostics previously read os.totalmem() directly
and now use getSystemTotalMb(), so container limits actually govern
parallel spawn decisions; CLI telemetry reports the effective total as
well. The cgroup probe result is cached for the process lifetime (the
limit is immutable per process) with a test reset hook; a detected limit
logs once so operators can see which source governs, and a
present-but-unreadable cgroup file warns once instead of failing
silently — absence stays silent. The root-path-vs-/proc/self/cgroup
trade-off is documented at the path constants. cli/tsconfig.json gains
the gcp-cloud-run/sdk source alias (matching the existing producer and
aws-lambda entries) so the cli typecheck resolves from source in a
fresh checkout.
Refs #1193, #1194, #1195, #1236
## What
Adds an auto-detected **low-memory safe render profile**. On hosts at or below 8 GB total RAM, the render pipeline collapses to its cheapest shape instead of running multiple concurrent Chrome instances.
When `lowMemoryMode` is active and the user hasn't passed `--workers`, the orchestrator:
- **skips auto-worker calibration** — no throwaway second Chrome just to time 5 frames;
- **pins to a single worker** — so the probe Chrome is reused for capture, never N concurrent;
- **prefers screenshot capture over BeginFrame** — avoids the BeginFrame protocol-timeout → relaunch churn on slow hardware;
- logs a one-line explanation of what it did and how to override.
Builds on #1221 (merged), which fixed the calibration timeout cap, the `<= 8192` boundary, and added the CLI timeout flags.
## Why
Reported in #1218 / #1219: renders on 8 GB laptops sit at low progress for minutes or stall. Root cause (per the triage thread) is architectural — the default pipeline launches up to 4 Chrome instances sequentially/overlapping (probe, calibration, capture, screenshot-fallback), each ~256 MB+, on machines with ~3 GB free. The concurrent browsers drive memory pressure that makes every CDP call slow and spikes V8 GC pauses.
#1221 made the timeouts and memory flags *apply correctly*; this PR removes the expensive shape entirely on the machines that can't afford it, rather than tuning it. "Smarter by default."
## How
- **`packages/engine/src/services/systemMemory.ts`** (new): one shared `isLowMemorySystem()` / `getSystemTotalMb()`, de-duplicating the `totalmem()` reads previously copied in `config.ts` and `browserManager.ts`. Threshold is inclusive (`<= 8192 MB`) — real "8 GB" hardware reports ~7600–8192 MB after firmware/iGPU reservations, so a strict `<` would skip the optimisation on the very hardware that needs it.
- **`config.ts`**: new `lowMemoryMode` field on `EngineConfig`, resolved tri-state — explicit override → `PRODUCER_LOW_MEMORY_MODE` (on/off) → auto-detect from total RAM.
- **`renderOrchestrator.ts`**: gate calibration off, pin workers to 1, force screenshot capture, and emit a safe-mode log line when `lowMemoryMode` is set and `--workers` is absent.
- **`render.ts`**: `--low-memory-mode` / `--no-low-memory-mode` override (sets the env var the producer's `resolveConfig` reads) + docs table entry.
Fully overridable: an explicit `--workers N` restores calibration-free parallelism; `--no-low-memory-mode` / `PRODUCER_LOW_MEMORY_MODE=false` restores the full default shape.
### Deliberately deferred (separate PRs)
- **Reuse the probe session for calibration**: only executes on the tier *above* 8 GB (safe-mode skips calibration on the target boxes). A correct BeginFrame-mode reuse would lose calibration's fast-fail-to-screenshot timeout — real risk on a path the reported scenario never hits. Better scoped on its own.
- **Retuning `calculateOptimalWorkers`'s `totalmem*0.5/256` memory model**: hot path for *all* renders incl. servers/Lambda, outside this PR's local-laptop scope.
## Test plan
- [x] Unit tests added/updated — `systemMemory.test.ts` (8192 boundary cases), `config.test.ts` (tri-state env resolution + explicit-override precedence). Engine suite passes (25 relevant tests).
- [x] `tsc` clean across engine/producer/cli; `oxlint` + `oxfmt` clean; removed an unused export so the `fallow --fail-on-issues` dead-code gate stays green.
- [x] Documentation updated — `docs/packages/cli.mdx` render-flags table.
- [ ] Manual testing on a real ≤ 8 GB host — not yet run; behaviour is unit-covered and the safe path (1 worker + screenshot) is already a supported render shape.
Note: one pre-existing producer test (`rejects a maliciously crafted key…`) fails identically on `main` — environment-specific path test, unrelated to this change.