[VaguelySerious]

Detection only. Nothing in the runtime branches on it yet. This is groundwork for a future run-ahead mode, where an inline step's successor starts before the step's step_completed write returns. Today's fast paths de-opt as soon as any hook or wait is open. This PR classifies, at each inline-step boundary, whether an event from an out-of-band writer could change what the boundary leads to.

What it detects

• Hooks: a hook counts as observed only while workflow code is waiting on its next payload (await hook, a pending for await iteration, .then). createHook registers a probe on the orchestrator context. The suspension snapshots it into WorkflowSuspension.observedHookIds when the VM goes idle. An unobserved hook's hook_received is buffered as an unarmed barrier and doesn't move the clock, so it can't change a decision until the first await.
• AbortController system hooks: always observed, because signal.aborted is read synchronously.
• Waits: sleep awaiters can't be seen. A wait counts as path-changing when its resumeAt falls inside the invocation's inline window plus clock skew (reuses waitDueThisInvocation). run.wakeUp() ignores resumeAt and is not covered; this is documented.
• Steps: any pending step this invocation doesn't execute itself counts as path-changing, because step awaiters are invisible.
• QuickJS: not covered. Its suspensions carry no observedHookIds, so every open hook counts as observed.

runtime/out-of-band-observation.ts documents the writer table, why unobserved hooks are inert, how a run-ahead gate should de-opt on this signal, and why run-ahead needs a pipeline-depth cap.

Output

observeOutOfBandWriters() returns an OutOfBandObservation (inert plus per-reason counts). It is computed in the inline batch gate in runtime.ts, only for batches that run a step inline. Per-invocation counters go on the span:

• workflow.out_of_band.inert_boundaries
• workflow.out_of_band.observed_hook_boundaries
• workflow.out_of_band.unknown_hook_boundaries (open hooks on a suspension without awaiter tracking, e.g. QuickJS)
• workflow.out_of_band.abort_signal_boundaries
• workflow.out_of_band.external_step_boundaries
• workflow.out_of_band.wait_due_boundaries

These measure how many boundaries run-ahead could pipeline.

Tests

out-of-band-observation.test.ts:

• Unit tests for the classifier.
• A sweep that replays five workflow shapes (late await, for await loop, background .then, hook raced against a step, and a sleep winning a race after an unawaited createHook). For every log position, it records whether the prefix reported the hook observed. It then inserts a payload there with a far-future createdAt and checks whether any step decision changes.
• Asserted: an insertion at an unobserved position never changes a decision. The observed cases do change, so the test catches a broken detector. Forcing the probe to false fails 4 of the tests.

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
example-nextjs-workflow-turbopack	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
example-nextjs-workflow-webpack	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
example-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-astro-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-express-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-fastify-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-hono-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-nestjs-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-nitro-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-nuxt-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-python-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-sveltekit-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-tanstack-start-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workbench-vite-workflow	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workflow-docs	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workflow-swc-playground	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workflow-tarballs	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC
workflow-web	Ready	Preview, v0	Oct 6, 2026 10:09pm UTC

[changeset-bot]

🦋 Changeset detected

Latest commit: 3e3a561

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 17 packages

Name	Type
@workflow/core	Patch
@workflow/builders	Patch
@workflow/cli	Patch
@workflow/next	Patch
@workflow/nitro	Patch
@workflow/vitest	Patch
@workflow/web-shared	Patch
@workflow/web	Patch
workflow	Patch
@workflow/world-testing	Patch
@workflow/astro	Patch
@workflow/nest	Patch
@workflow/rollup	Patch
@workflow/sveltekit	Patch
@workflow/vite	Patch
@workflow/nuxt	Patch
@workflow/world-vercel	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[github-actions]

📊 Workflow Benchmarks

commit 3e3a561 · Tue, 06 Oct 2026 22:32:07 GMT · run logs

Backend: vercel · app: nextjs-turbopack

Metric	Scenario	Best (ms)	P75 (ms)	P90 (ms)	P99 (ms)	Samples
TTFS	step	345	2197 🔴	2309 🔴	2447 🔴	30
TTFS	stream	358	2316 🔴	2371 🔴	2651 🔴	30
TTFS	hook + stream	2458	2572 🔴	2606 🔴	2676 🔴	30
Fan-out TTFS	Promise.all(100 steps)	534	848	981	2686	10
Fan-out TTLS	Promise.all(100 steps)	2948	4491	4842	6229	10
STSO	1020 steps (inline)	118	155	178	329	1019
WO	1020 steps	154789	154789	154789	154789	1
CRTT	first chunk (pooled)	30	72	110	171	28

Streams

Scenario	CRTT 1st	p75	p90	p99	CDV max	iters
paced control (100/s, 60B)	40.5	81	135	350	124	10
size sweep (100/s, 160B-12KB)	44.5	89	162	335	144	10
replay gateway-gpt-5.4-nano-2000t (1x)	51	77	113	309	165	3
replay eve-gpt-5.6-sol-2000t (1x)	69	75	103	245	220	2
replay eve-gpt-5.6-sol-2000t (2x)	45	91	146	1140	160	3

📈 STSO distribution (inline / queue-hop histograms)

1020 steps (inline)

Cumulative STSO time: 154408ms over 1019 samples

_{No main baseline with raw samples yet — showing this run's distribution on its own; the diff appears once a run on main has recorded them.}

100-150 ms  ████████████████████████  steps 688
150-200 ms  ██████████                steps 283
200-250 ms  █                         steps  32
250-300 ms  █                         steps   4
300-350 ms  █                         steps   2
350-400 ms  █                         steps   3
400-450 ms  █                         steps   1
450-500 ms  █                         steps   1
500-550 ms  █                         steps   2
550-600 ms  █                         steps   1
600-650 ms  █                         steps   1
900-950 ms  █                         steps   1

📈 CRTT drill-down (RTT distributions & profiles)

_{No main baseline yet — percentages appear once a run on main has recorded CRTT.}

variant  RTT 1ms→5s+     avg  p50  p90   p99     n
control  ·····▃█▁▁····  69.8   60  135   350  3000
sweep    ·····▃█▂▁····  74.3   64  162   335  3000
gw 1x    ·····▅█▂▁····  64.5   55  113   309  5295
eve 1x   ·····▅█▂▁····  63.8   56  103   245  5186
eve 2x   ·····▂█▂▁▁▁··    92   68  146  1140  7779

RTT over stream progress (avg per tenth of stream, bars scaled min→max):

control  ▇▄▃▁▄▆█▃▂▃  60–82ms
sweep    ▅█▄▅██▂▇▄▁  63–82ms
gw 1x    ▆█▃▃▃▃▁▃▄▄  53–83ms
eve 1x   ▄▅▁▂▄▂▅█▇▇  59–69ms
eve 2x   ▁▁▁▁▁▁▂█▁▁  67–258ms

RTT by chunk size (avg per log size bin, ~160B → ~12KB serialized, bars scaled min→max):

sweep  ▇██▄▅▁▁  72–76ms

Delivery jitter over stream progress (avg positive CDV per tenth of stream, bars scaled min→max):

control  █▄▂▁▄▂▆▃▂▄  14–20ms
sweep    ▄▄▃▁█▃▂▅▂▁  19–28ms
gw 1x    ▇█▃▂▄▄▁▃▄▅  10–17ms
eve 1x   ▂▃▁▃▄▄▁▄▅█  14–21ms
eve 2x   ▂▂▁▂▂▂▅█▂▄  13–24ms

📜 Previous results (1)

d086b1e

Tue, 06 Oct 2026 19:53:34 GMT · run logs

vercel / nextjs-turbopack

Metric	Scenario	Best (ms)	P75 (ms)	P90 (ms)	P99 (ms)	Samples
TTFS	step	422 (-42%) 💚	2240 🔴 (+105%) 🔻	2251 🔴 (+102%) 🔻	2348 🔴 (+95%) 🔻	30
TTFS	stream	1945 (+784%) 🔻	2182 🔴 (+108%) 🔻	2226 🔴 (+103%) 🔻	2454 🔴 (+72%) 🔻	30
TTFS	hook + stream	618 (-30%) 💚	2455 🔴 (+85%) 🔻	2513 🔴 (+79%) 🔻	2753 🔴 (+76%) 🔻	30
Fan-out TTFS	Promise.all(100 steps)	518 (-40%) 💚	683 (-60%) 💚	698 (-61%) 💚	2558 (+37%) 🔻	10
Fan-out TTLS	Promise.all(100 steps)	1803 (-21%) 💚	3031 (-17%) 💚	3044 (-19%) 💚	8233 (+14%)	10
STSO	1020 steps (inline)	104 (+4.0%)	143 (+10%)	159 (+3.2%)	233 (+15%)	1019
WO	1020 steps	144126 (+9.4%)	144126 (+9.4%)	144126 (+9.4%)	144126 (+9.4%)	1
CRTT	first chunk (pooled)	25 (-65%) 💚	46 (-68%) 💚	93 (-56%) 💚	146 (-59%) 💚	28

Streams

Scenario	CRTT 1st	p75	p90	p99	CDV max	iters
paced control (100/s, 60B)	33.5 (-66%)	69 (-64%)	120 (-60%)	335 (-32%)	89 (-60%)	10
size sweep (100/s, 160B-12KB)	32 (-69%)	65 (-71%)	122 (-65%)	559 (-12%)	120 (-43%)	10
replay gateway-gpt-5.4-nano-2000t (1x)	34 (-75%)	63 (-67%)	153 (-52%)	2438 (+325%)	546 (+25%)	3
replay eve-gpt-5.6-sol-2000t (1x)	45.5 (-56%)	72 (-63%)	105 (-66%)	270 (-51%)	290 (-36%)	2
replay eve-gpt-5.6-sol-2000t (2x)	41 (-71%)	71 (-78%)	130 (-73%)	438 (-53%)	286 (-39%)	3

ℹ️ Metric definitions & methodology

_{Streams: first-chunk RTT (the stream-open path, before any buffering/backpressure), CRTT percentiles, and worst delivery stall (CDV max). Cells are medians across iterations; per-run values in the artifacts. No 🔴/🟢 marks until targets attach.}

_{The collapsed STSO distribution section above buckets every step gap, split inline (same warm process — pure framework overhead) vs queue-hop (fresh process — dispatch, reinit, replay). █ = main, ┃ = this run, ░ = fill.}

_{The collapsed CRTT drill-down: per-variant RTT histograms (fixed log bins, · = empty) and mean RTT/positive-CDV profile lines over stream progress and chunk size. Histograms, avgs, and profiles merge exactly across runs; p50–p99 are percentile-of-percentiles. Per-index rows live in the artifacts.}

_{Metrics — TTFS: time to first step body (in-deployment start() → first step body) · Fan-out TTFS: fan-out time to first step (in-deployment start() → first of the parallel step bodies to complete) · Fan-out TTLS: fan-out time to last step (in-deployment start() → last of the parallel step bodies to complete, i.e. when the Promise.all resolves) · STSO: step-to-step overhead (gap between consecutive step bodies) · WO: workflow overhead (whole-run time outside step bodies, in-deployment anchored) · CRTT: chunk round-trip time (per-chunk write → read latency, one clock domain: deployment → stream backend → same deployment) · CDV: chunk delay variation / delivery jitter (inter-arrival gap minus inter-write gap per seq-adjacent pair; skew-free; the row is each run's MAX positive value, so one stall moves it)}

_{Scenarios — step: one trivial no-op step, no stream; no hooks, so the run stays in turbo mode (in-process fast path) · stream: one streaming step; no hooks, so the run stays in turbo mode (in-process fast path) · hook + stream: registers a hook before one step, which exits turbo mode (dispatch path) · 1020 steps: 1020 trivial sequential steps; STSO is measured between consecutive steps in the given step ranges, and WO is the whole-run overhead outside step bodies · Promise.all(100 steps): 100 trivial no-op steps started together in a single Promise.all; Fan-out TTFS is the first of them to complete and Fan-out TTLS the last, both from the in-deployment clientStart, so their gap is the spread the runtime adds across the fan-out · paced control (100/s, 60B): the control: 300 tiny (~60B) deltas metronome-paced at 100/s — zero workload structure, so it reads the transport floor and flush cadence, and disambiguates transport-wide vs workload-specific when a replay row moves · size sweep (100/s, 160B-12KB): same pacing as the control with deltas padded in rotation across seven log-spaced sizes (~160B–12KB) — rotation decouples size from stream position, so it isolates whether chunk size causes latency · replay gateway-gpt-5.4-nano-2000t (1x): raw provider SSE cadence captured at the AI gateway boundary (gpt-5.4-nano, the most popular gateway model; per-token deltas p50 208B = the modal production chunk size), replayed exactly as measured — the typical customer's workload; its CDV is the typical customer's real delivery jitter · replay eve-gpt-5.6-sol-2000t (1x): a captured eve turn (gpt-5.6-sol, the most-used demanding eve model; ~2000 output tokens = production p50 turn length) replayed exactly as measured — eve's envelope protocol re-ships the cumulative message so sizes ramp 142B→13KB; the demanding outlier tenant's reality · replay eve-gpt-5.6-sol-2000t (2x): the same eve capture at 2x — the headroom/stress row; real fast-tier models emit the same chunk sizes at proportionally higher rate, so time compression is a faithful speed model · first chunk (pooled): every run's seq-0 RTT pooled across all stream scenarios — the first chunk precedes any workload differentiation, so pooling samples one shared stream-open path with exact percentiles}

_{Replay cadences (semantic sha256) — eve-gpt-5.6-sol-2000t eaf22f5946e7c61f3c65c7006d550df180cfabd4e706254a09f22aec0cfb420d · gateway-gpt-5.4-nano-2000t 6f24ac518b6b83ff1d0e85a5fe78230db192716d66a7fc6b2fe022752001d041}

_{🔴 marks a percentile over its target (within target is left unmarked). Targets (p75/p90/p99, ms) — TTFS 200/300/600}

_{All timestamps are deployment-side; runs are triggered in-deployment, so the CI runner and api.vercel.com sit outside every measured window. TTFS = start() → first step body (includes dispatch + any cold start); Fan-out TTFS/TTLS = first/last step completion of one Promise.all from the same anchor (the gap is the runtime’s fan-out spread); STSO/WO between step bodies; CRTT inside the workflow (excludes the api.vercel.com read path).}

_{Cold starts stay in the numbers (real bursty-workload latency, inflates P75+); Best is the warm floor.}

[github-actions]

🧪 E2E Test Results

✅ All tests passed

⚠️ Flaky E2E Tests (passed on retry)

These tests failed at least once and passed on a retry. A recurring entry here is a real race worth investigating.

• a chain of takeovers: each victim ends with HookForceClaimedError, deliveries follow the current owner (nextjs-webpack · local-dev / local / node / stable)
• a run can take over its own earlier hook (sveltekit · local-dev / local / node / stable)

🛠 Infra Events (absorbed by the harness)

Platform anomalies the e2e harness detected and worked around (e.g. a run the queue never picked up, replaced by a fresh run). Clustered timestamps indicate a backend blip; a steady drip indicates a platform issue worth escalating.

• run-pickup-stall · runs app-generated source against a registered step (tanstack-start) · at 22:11:05Z · abandoned wrun_01M49M9YY5XX91HGNTWK9FTG3C
• cold-start-warmup · suite warmup (tanstack-start) · at 22:11:08Z · abandoned wrun_01M49M9YY3KVWR131F8XDQC836
• run-pickup-stall · workflow throw of a non-Error value round-trips verbatim as cause (nuxt) · at 22:12:02Z · abandoned wrun_01M49MBQ2A0HC0NQ0EBEZZ8HYC
• run-pickup-stall · hookCleanupTestWorkflow - hook token reuse after workflow completion (nextjs-webpack) · at 22:17:12Z · abandoned wrun_01M49MN55AEMHKQRQYJ19JT89H

E2E Test Summary

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	3904	0	875	4779
✅ 💻 Local Development	4582	0	551	5133
✅ 📦 Local Production	4582	0	551	5133
✅ 🐘 Local Postgres	4582	0	551	5133
✅ 🪟 Windows	342	0	12	354
✅ 🌐 Cross-language Conformance	68	0	84	152
✅ dynamic-runs	0	0	0	0
✅ vercel-http-transport	879	0	183	1062
✅ vercel-multi-region	27	0	0	27
✅ vercel-ws-transport	595	0	113	708
Total	19561	0	2920	22481

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro-node	142	0	35
✅ astro-quickjs	142	0	35
✅ example-node	142	0	35
✅ example-quickjs	142	0	35
✅ express-node	142	0	35
✅ express-quickjs	142	0	35
✅ fastify-node	142	0	35
✅ fastify-quickjs	142	0	35
✅ hono-node	142	0	35
✅ hono-quickjs	142	0	35
✅ nest-node	142	0	35
✅ nest-quickjs	142	0	35
✅ nextjs-turbopack-node	169	0	8
✅ nextjs-turbopack-quickjs	169	0	8
✅ nextjs-webpack-node	169	0	8
✅ nextjs-webpack-quickjs	169	0	8
✅ nitro-node	142	0	35
✅ nitro-quickjs	142	0	35
✅ nuxt-node	142	0	35
✅ nuxt-quickjs	142	0	35
✅ python-node	66	0	111
✅ sveltekit-node	161	0	16
✅ sveltekit-quickjs	161	0	16
✅ tanstack-start-node	142	0	35
✅ tanstack-start-quickjs	142	0	35
✅ vite-node	142	0	35
✅ vite-quickjs	142	0	35

✅ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable-node	148	0	29
✅ astro-stable-quickjs	148	0	29
✅ express-stable-node	148	0	29
✅ express-stable-quickjs	148	0	29
✅ fastify-stable-node	148	0	29
✅ fastify-stable-quickjs	148	0	29
✅ hono-stable-node	148	0	29
✅ hono-stable-quickjs	148	0	29
✅ nest-stable-node	148	0	29
✅ nest-stable-quickjs	148	0	29
✅ nextjs-turbopack-canary-node	176	0	1
✅ nextjs-turbopack-canary-quickjs	176	0	1
✅ nextjs-turbopack-quickjs-snapshot	176	0	1
✅ nextjs-turbopack-stable-node	176	0	1
✅ nextjs-turbopack-stable-quickjs	176	0	1
✅ nextjs-webpack-canary-node	176	0	1
✅ nextjs-webpack-canary-quickjs	176	0	1
✅ nextjs-webpack-stable-node	176	0	1
✅ nextjs-webpack-stable-quickjs	176	0	1
✅ nitro-stable-node	148	0	29
✅ nitro-stable-quickjs	148	0	29
✅ nuxt-stable-node	148	0	29
✅ nuxt-stable-quickjs	148	0	29
✅ sveltekit-stable-node	167	0	10
✅ sveltekit-stable-quickjs	167	0	10
✅ tanstack-start-node	148	0	29
✅ tanstack-start-quickjs	148	0	29
✅ vite-stable-node	148	0	29
✅ vite-stable-quickjs	148	0	29

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable-node	148	0	29
✅ astro-stable-quickjs	148	0	29
✅ express-stable-node	148	0	29
✅ express-stable-quickjs	148	0	29
✅ fastify-stable-node	148	0	29
✅ fastify-stable-quickjs	148	0	29
✅ hono-stable-node	148	0	29
✅ hono-stable-quickjs	148	0	29
✅ nest-stable-node	148	0	29
✅ nest-stable-quickjs	148	0	29
✅ nextjs-turbopack-canary-node	176	0	1
✅ nextjs-turbopack-canary-quickjs	176	0	1
✅ nextjs-turbopack-quickjs-snapshot	176	0	1
✅ nextjs-turbopack-stable-node	176	0	1
✅ nextjs-turbopack-stable-quickjs	176	0	1
✅ nextjs-webpack-canary-node	176	0	1
✅ nextjs-webpack-canary-quickjs	176	0	1
✅ nextjs-webpack-stable-node	176	0	1
✅ nextjs-webpack-stable-quickjs	176	0	1
✅ nitro-stable-node	148	0	29
✅ nitro-stable-quickjs	148	0	29
✅ nuxt-stable-node	148	0	29
✅ nuxt-stable-quickjs	148	0	29
✅ sveltekit-stable-node	167	0	10
✅ sveltekit-stable-quickjs	167	0	10
✅ tanstack-start-node	148	0	29
✅ tanstack-start-quickjs	148	0	29
✅ vite-stable-node	148	0	29
✅ vite-stable-quickjs	148	0	29

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable-node	148	0	29
✅ astro-stable-quickjs	148	0	29
✅ express-stable-node	148	0	29
✅ express-stable-quickjs	148	0	29
✅ fastify-stable-node	148	0	29
✅ fastify-stable-quickjs	148	0	29
✅ hono-stable-node	148	0	29
✅ hono-stable-quickjs	148	0	29
✅ nest-stable-node	148	0	29
✅ nest-stable-quickjs	148	0	29
✅ nextjs-turbopack-canary-node	176	0	1
✅ nextjs-turbopack-canary-quickjs	176	0	1
✅ nextjs-turbopack-quickjs-snapshot	176	0	1
✅ nextjs-turbopack-stable-node	176	0	1
✅ nextjs-turbopack-stable-quickjs	176	0	1
✅ nextjs-webpack-canary-node	176	0	1
✅ nextjs-webpack-canary-quickjs	176	0	1
✅ nextjs-webpack-stable-node	176	0	1
✅ nextjs-webpack-stable-quickjs	176	0	1
✅ nitro-stable-node	148	0	29
✅ nitro-stable-quickjs	148	0	29
✅ nuxt-stable-node	148	0	29
✅ nuxt-stable-quickjs	148	0	29
✅ sveltekit-stable-node	167	0	10
✅ sveltekit-stable-quickjs	167	0	10
✅ tanstack-start-node	148	0	29
✅ tanstack-start-quickjs	148	0	29
✅ vite-stable-node	148	0	29
✅ vite-stable-quickjs	148	0	29

✅ 🪟 Windows

App	Passed	Failed	Skipped
✅ nextjs-turbopack-node	171	0	6
✅ nextjs-turbopack-quickjs	171	0	6

✅ 🌐 Cross-language Conformance

App	Passed	Failed	Skipped
✅ python	68	0	84

✅ dynamic-runs

App	Passed	Failed	Skipped
✅ astro-vercel	0	0	0
✅ example-vercel	0	0	0
✅ express-vercel	0	0	0
✅ fastify-vercel	0	0	0
✅ hono-vercel	0	0	0
✅ nest-vercel	0	0	0
✅ nextjs-turbopack-vercel	0	0	0
✅ nextjs-webpack-vercel	0	0	0
✅ nitro-vercel	0	0	0
✅ nuxt-vercel	0	0	0
✅ sveltekit-vercel	0	0	0
✅ tanstack-start-vercel	0	0	0
✅ vite-vercel	0	0	0

✅ vercel-http-transport

App	Passed	Failed	Skipped
✅ example	142	0	35
✅ express	142	0	35
✅ hono	142	0	35
✅ nextjs-turbopack	169	0	8
✅ nitro	142	0	35
✅ vite	142	0	35

✅ vercel-multi-region

App	Passed	Failed	Skipped
✅ nextjs-turbopack	27	0	0

✅ vercel-ws-transport

App	Passed	Failed	Skipped
✅ example	142	0	35
✅ express	142	0	35
✅ nextjs-turbopack	169	0	8
✅ vite	142	0	35

---

📋 View full workflow run

[github-actions]

Sim World

Simulated world deterministic testing for races. Traces

🟠 world-sim scenario book — 1 fail of 42 total

fence=per-spec

scenario	outcome	events	virt	replay	violations
✅ smoke-no-steps	completed	3	0ms	ok	0
✅ smoke-one-step	completed	6	0ms	ok	0
✅ hook-at-step-started	completed	12	0ms	ok	0
✅ hook-at-step-completed	completed	12	0ms	ok	0
✅ hook-at-hook-created	completed	12	0ms	ok	0
✅ deadline-hook-wins	completed	7	1.0h	ok	0
✅ deadline-expires	completed	7	1.0h	ok	0
✅ step-vs-timer-early-settlement	completed	8	1.0h	ok	0
✅ long-sleep	completed	11	30.0d	ok	0
✅ hook-never-arrives	stalled	3	0ms	skipped	0
✅ step-retries-twice	completed	10	2.0s	ok	0
✅ parallel-steps	completed	9	0ms	ok	0
✅ hook-on-execution-state	completed	12	0ms	ok	0
✅ peek-hook-before-branch	completed	12	0ms	ok	0
✅ peek-hook-after-branch	completed	12	0ms	ok	0
✅ peek-hook-at-registration	completed	12	0ms	ok	0
✅ race-hook-before-probe	completed	12	0ms	ok	0
✅ race-hook-after-probe	completed	12	0ms	ok	0
✅ race-duplicate-delivery	completed	13	0ms	ok	0
✅ attr-hook-before-step	completed	11	0ms	ok	0
✅ attr-hook-after-step	completed	11	0ms	ok	0
✅ attr-from-step-body	completed	13	0ms	ok	0
✅ fork-hook-after-timeout	completed	14	1.0m	ok	0
✅ fork-hook-before-timeout	completed	14	1.0m	ok	0
✅ count-hook-after-timeout	completed	17	1.0m	ok	0
✅ count-hook-before-timeout	completed	20	1.0m	ok	0
✅ stale-read-step-count-fork	completed	20	1.0m	ok	0
✅ stale-read-equal-step-counts	completed	14	1.0m	ok	0
✅ step-vs-step-fork	completed	12	0ms	ok	0
✅ step-vs-step-fork-fenced	completed	12	0ms	ok	0
✅ fence-catches-benign-direction	completed	12	5ms	ok	0
✅ in-flight-before-decision	completed	17	1.0m	ok	0
❌ in-flight-before-decision-counted	completed	17	1.0m	ok	0
✅ in-flight-after-decision	completed	19	2.0m	ok	0
✅ stale-read-step-count-fork-fenced	completed	20	1.0m	ok	0
✅ fork-hook-wins	completed	13	1.0m	ok	0
✅ fork-timeout-wins	completed	13	1.0m	ok	0
✅ unclaimed-payload-under-fork	completed	17	1.0m	ok	0
✅ claimed-payload-under-fork	completed	17	1.0m	ok	0
✅ writers-independent-step-bodies	completed	12	0ms	ok	0
✅ writers-scripted-tempo	completed	12	0ms	ok	0
✅ cancel-mid-step	cancelled	7	0ms	skipped	0

Full trace: world-sim.txt

[github-actions]

Framework	Flow route	Step reg.	Framework output
hono	267.4 KiB (±0)	97.4 KiB (-38 B)	2.01 MiB (+563 B)
nextjs-turbopack	274.8 KiB (±0)	426 B (±0)	987.8 KiB (+295 B)

About these numbers

Sizes are gzip; parentheses show the change against main.
Flow route and Step reg. gate this job, on raw bytes rather than the gzip shown, at max(2%, 50.0 KiB). Framework output is informational.

3e3a561 · run

[TooTallNate]

AI review: approve. I found nothing blocking, so there are no inline threads.

What I checked:

• Probe correctness (hook.ts). promises.length > 0 tracks exactly the state that decides whether a hook_received registers an armed or unarmed barrier (hasWaitingConsumer uses the same expression). So "observed" and "this payload would be ordered against step results" are the same predicate and can't drift. Buffered-claim, for await between iterations, conflict and force-claim drains, and the Promise.race leftover over-report all behave as the module doc describes.
• Ordering argument. A later wait_completed or registration delivery does gate on an unarmed hook barrier (DEFER_BEHIND). A step that follows it still gates on that armed entry, because the unarmed skip is direct, never transitive. Log order is preserved, so the "unobserved payload is inert" claim holds for the wait/registration interleavings I traced as well, not only the four swept shapes.
• Snapshot timing. observedHookIds is taken inside the same scheduleWhenIdle callback that builds the suspension, so the probe and the queue describe the same instant. The completion-drain suspension in workflow.ts doesn't reach the batch gate. QuickJS degrades to unknownHookCount, which is conservative.
• runtime.ts wiring. inlineCorrelationIds, ownedRecoverySteps and waitDueThisInvocation are all populated before the call. Backstopped and queued steps correctly count as external. Nothing branches on the result.
• I ran it locally: out-of-band-observation.test.ts passes (11/11) and tsc --noEmit is clean for @workflow/core. CI is green.

Non-blocking suggestions, all fine as follow-ups:

1. observed_hook_boundaries merges observedHookCount and unknownHookCount. If QuickJS runs reach this gate, every boundary with an open hook shows up as "observed hook" and skews the metric this PR is meant to collect. A separate unknown_hook_boundaries counter (or a workflow.out_of_band.hook_tracking: false attribute) would keep the measurement honest.
2. observeOutOfBandWriters(...) and its new Set([...]) run on every batch suspension, but the result is only used when lazyInlineSteps.length > 0. Moving the call inside that branch avoids the allocation on the queue-only path.
3. hookPayloadAwaiters entries are only removed on hook_conflict and hook_disposed events, not on a local dispose(). This is harmless today: the item is skipped via disposed, and the map is per-session. Dropping the entry in disposeHook too would match the "dropped once the hook can receive nothing more" doc on the field.
4. The sweep only compares step decisions (name and args). Adding one shape with a sleep raced against a step after an unawaited createHook would pin the wait-barrier interleaving above with a test, which is the case most likely to regress once run-ahead starts relying on inert.

[VaguelySerious]

(AI) Addressed all four suggestions from the review in 3e3a561:

1. Boundaries whose suspension has no awaiter tracking now count under a separate workflow.out_of_band.unknown_hook_boundaries instead of observed_hook_boundaries.
2. observeOutOfBandWriters only runs when the batch has an inline step. outOfBandObservation is undefined otherwise.
3. disposeHook drops the hook's entry from hookPayloadAwaiters.
4. New sweep shape: an unawaited createHook, then sleep raced against a step with the sleep winning, so the wait_completed is delivered behind an unarmed hook barrier wherever the payload is inserted. The test asserts the sleep wins in the baseline, the hook stays unobserved until its await, and no insertion changes a step decision.

[github-actions]

No backport to stable for 3949c07 (AI decision).

This commit adds new detection machinery (observeOutOfBandWriters, WorkflowSuspension.observedHookIds, hookPayloadAwaiters, and six new span attributes) explicitly as groundwork for a future run-ahead mode — the PR states nothing in the runtime branches on it yet. It fixes no user-visible defect, so it is additive feature/instrumentation work rather than a stability fix, and it does not belong on the maintenance line.

To override, re-run the Backport to stable workflow manually via workflow_dispatch and paste this commit SHA into the ref input:

3949c0717bbff96d2527ef7b5397e98684974f07