ai sdk parity for DurableAgent

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example

[ctate]

Key Changes

Generation Settings

• Support for temperature, maxOutputTokens, topP, topK, presencePenalty, frequencyPenalty, stopSequences, seed at constructor and per-stream level

Step Control

• Added maxSteps to limit LLM calls (default: unlimited for backwards compatibility)
• Added toolChoice (auto, none, required, or specific tool)
• Added activeTools to filter available tools per-stream call

Callbacks

• onChunk - called for each stream chunk
• onStepFinish - called after each step completes
• onFinish - called when all steps complete
• onError - called on errors
• onAbort - called when aborted via AbortSignal

Structured Output

• Added experimental_output with Output.object({ schema }) for parsing structured responses

Context & Telemetry

• Added experimental_context to pass data to tool executions
• Added experimental_telemetry for observability

Stream Result

• stream() now returns { messages, steps, experimental_output } instead of just { messages }

Other

• prepareStep can now override generation settings, toolChoice, and context
• Added experimental_repairToolCall for handling malformed tool calls
• Enabled file, source, and tool input streaming chunks
• Exported all new types from @workflow/ai

[vercel]

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

Project	Deployment	Review	Updated (UTC)
example-nextjs-workflow-turbopack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-nextjs-workflow-webpack	Ready	Preview, Comment	Dec 24, 2025 5:31pm
example-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-astro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-express-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-fastify-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-hono-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nitro-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-nuxt-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-sveltekit-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workbench-vite-workflow	Ready	Preview, Comment	Dec 24, 2025 5:31pm
workflow-docs	Ready	Preview, Comment	Dec 24, 2025 5:31pm

[changeset-bot]

🦋 Changeset detected

Latest commit: 0c0dfe1

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

This PR includes changesets to release 1 package

Name	Type
@workflow/ai	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]

🧪 E2E Test Results

❌ Some tests failed

Summary

	Passed	Failed	Skipped	Total
✅ ▲ Vercel Production	286	0	11	297
❌ 💻 Local Development	236	26	8	270
✅ 📦 Local Production	262	0	8	270
✅ 🐘 Local Postgres	262	0	8	270
❌ 🪟 Windows	0	27	0	27
❌ 🌍 Community Worlds	58	62	0	120
Total	1104	115	35	1254

❌ Failed Tests

💻 Local Development (26 failed)

nuxt-stable (26 failed):

• addTenWorkflow
• addTenWorkflow
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🪟 Windows (27 failed)

nextjs-turbopack (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

🌍 Community Worlds (62 failed)

mongodb (1 failed):

• webhookWorkflow

redis (1 failed):

• webhookWorkflow

starter-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

starter (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

turso-dev (3 failed):

• dev e2e should rebuild on step change
• dev e2e should rebuild on workflow change
• dev e2e should rebuild on adding workflow file

turso (27 failed):

• addTenWorkflow
• addTenWorkflow
• should work with react rendering in step
• promiseAllWorkflow
• promiseRaceWorkflow
• promiseAnyWorkflow
• readableStreamWorkflow
• hookWorkflow
• webhookWorkflow
• webhook route with invalid token
• sleepingWorkflow
• nullByteWorkflow
• workflowAndStepMetadataWorkflow
• outputStreamWorkflow
• outputStreamInsideStepWorkflow - getWritable() called inside step functions
• fetchWorkflow
• promiseRaceStressTestWorkflow
• retryAttemptCounterWorkflow
• retryableAndFatalErrorWorkflow
• stepDirectCallWorkflow - calling step functions directly outside workflow context
• crossFileErrorWorkflow - stack traces work across imported modules
• hookCleanupTestWorkflow - hook token reuse after workflow completion
• stepFunctionPassingWorkflow - step function references can be passed as arguments (without closure vars)
• stepFunctionWithClosureWorkflow - step function with closure variables passed as argument
• closureVariableWorkflow - nested step functions with closure variables
• spawnWorkflowFromStepWorkflow - spawning a child workflow using start() inside a step
• health check endpoint - workflow and step endpoints respond to __health query parameter

Details by Category

✅ ▲ Vercel Production

App	Passed	Failed	Skipped
✅ astro	26	0	1
✅ example	26	0	1
✅ express	26	0	1
✅ fastify	26	0	1
✅ hono	26	0	1
✅ nextjs-turbopack	26	0	1
✅ nextjs-webpack	26	0	1
✅ nitro	26	0	1
✅ nuxt	26	0	1
✅ sveltekit	26	0	1
✅ vite	26	0	1

❌ 💻 Local Development

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
❌ nuxt-stable	0	26	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 📦 Local Production

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

✅ 🐘 Local Postgres

App	Passed	Failed	Skipped
✅ astro-stable	26	0	1
✅ express-stable	26	0	1
✅ fastify-stable	26	0	1
✅ hono-stable	26	0	1
✅ nextjs-turbopack-stable	27	0	0
✅ nextjs-webpack-stable	27	0	0
✅ nitro-stable	26	0	1
✅ nuxt-stable	26	0	1
✅ sveltekit-stable	26	0	1
✅ vite-stable	26	0	1

❌ 🪟 Windows

App	Passed	Failed	Skipped
❌ nextjs-turbopack	0	27	0

❌ 🌍 Community Worlds

App	Passed	Failed	Skipped
✅ mongodb-dev	3	0	0
❌ mongodb	26	1	0
✅ redis-dev	3	0	0
❌ redis	26	1	0
❌ starter-dev	0	3	0
❌ starter	0	27	0
❌ turso-dev	0	3	0
❌ turso	0	27	0

---

📋 View full workflow run

---

❌ Some E2E test jobs failed:

• Vercel Prod: success
• Local Dev: failure
• Local Prod: success
• Local Postgres: success
• Windows: success

Check the workflow run for details.

[github-actions]

📊 Benchmark Results

📈 Comparing against baseline from main branch. Green 🟢 = faster, Red 🔺 = slower.

workflow with no steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.031s (-26.1% 🟢)	1.007s (~)	0.975s	10	1.00x
🌐 Starter	Next.js (Turbopack)	0.036s (-8.7% 🟢)	1.014s (~)	0.978s	10	1.15x
🌐 Redis	Next.js (Turbopack)	0.037s (-8.2% 🟢)	1.018s (~)	0.982s	10	1.18x
💻 Local	Next.js (Turbopack)	0.038s (-4.5%)	1.018s (~)	0.980s	10	1.21x
💻 Local	Express	0.042s (-5.8% 🟢)	1.007s (~)	0.965s	10	1.35x
🌐 Turso	Next.js (Turbopack)	0.104s (-3.3%)	1.014s (~)	0.910s	10	3.34x
🌐 MongoDB	Next.js (Turbopack)	0.110s (-12.9% 🟢)	1.014s (~)	0.904s	10	3.52x
🐘 Postgres	Nitro	0.201s (-11.7% 🟢)	1.027s (+0.7%)	0.827s	10	6.43x
🐘 Postgres	Next.js (Turbopack)	0.247s (+11.7% 🔺)	1.020s (~)	0.773s	10	7.92x
🐘 Postgres	Express	0.380s (+28.2% 🔺)	1.013s (~)	0.633s	10	12.19x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	0.569s (-3.6%)	1.560s (-2.5%)	0.991s	10	1.00x
▲ Vercel	Next.js (Turbopack)	0.588s (-7.3% 🟢)	1.642s (+10.5% 🔺)	1.054s	10	1.03x
▲ Vercel	Nitro	0.649s (+29.2% 🔺)	1.603s (+6.1% 🔺)	0.954s	10	1.14x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

workflow with 1 step

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.078s (-3.2%)	2.005s (~)	0.927s	10	1.00x
🌐 Redis	Next.js (Turbopack)	1.082s (-1.6%)	2.015s (~)	0.933s	10	1.00x
💻 Local	Next.js (Turbopack)	1.095s (~)	2.011s (~)	0.917s	10	1.02x
🌐 Starter	Next.js (Turbopack)	1.095s (~)	2.009s (~)	0.915s	10	1.02x
💻 Local	Express	1.112s (~)	2.006s (~)	0.894s	10	1.03x
🌐 Turso	Next.js (Turbopack)	1.306s (~)	2.013s (~)	0.706s	10	1.21x
🌐 MongoDB	Next.js (Turbopack)	1.311s (~)	2.013s (~)	0.702s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	1.817s (-4.5%)	2.015s (~)	0.198s	10	1.69x
🐘 Postgres	Express	2.237s (+5.4% 🔺)	3.015s (~)	0.778s	10	2.08x
🐘 Postgres	Nitro	2.450s (+7.3% 🔺)	3.014s (~)	0.565s	10	2.27x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.680s (~)	3.564s (-1.5%)	0.884s	10	1.00x
▲ Vercel	Next.js (Turbopack)	2.720s (-1.7%)	3.807s (+3.9%)	1.087s	10	1.01x
▲ Vercel	Express	2.915s (-9.0% 🟢)	3.798s (-8.9% 🟢)	0.883s	10	1.09x

🔍 Observability: Nitro | Next.js (Turbopack) | Express

workflow with 10 sequential steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	10.530s (-2.3%)	11.011s (~)	0.481s	5	1.00x
🌐 Redis	Next.js (Turbopack)	10.592s (~)	11.024s (~)	0.432s	5	1.01x
🌐 Starter	Next.js (Turbopack)	10.593s (~)	11.010s (~)	0.418s	5	1.01x
💻 Local	Next.js (Turbopack)	10.655s (~)	11.016s (~)	0.361s	5	1.01x
💻 Local	Express	10.788s (~)	11.013s (~)	0.225s	5	1.02x
🌐 Turso	Next.js (Turbopack)	12.184s (~)	13.021s (~)	0.837s	5	1.16x
🌐 MongoDB	Next.js (Turbopack)	12.220s (~)	13.021s (~)	0.801s	5	1.16x
🐘 Postgres	Next.js (Turbopack)	15.222s (+1.1%)	16.037s (+2.6%)	0.816s	5	1.45x
🐘 Postgres	Express	16.138s (-21.1% 🟢)	16.831s (-20.0% 🟢)	0.693s	5	1.53x
🐘 Postgres	Nitro	20.335s (~)	21.039s (~)	0.704s	5	1.93x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	21.391s (+1.5%)	22.091s (~)	0.700s	5	1.00x
▲ Vercel	Next.js (Turbopack)	21.549s (+0.8%)	22.240s (~)	0.691s	5	1.01x
▲ Vercel	Nitro	22.032s (~)	22.680s (-1.1%)	0.647s	5	1.03x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.all with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.329s (-5.3% 🟢)	2.006s (~)	0.678s	15	1.00x
🌐 Starter	Next.js (Turbopack)	1.344s (~)	2.009s (~)	0.665s	15	1.01x
🌐 Redis	Next.js (Turbopack)	1.356s (~)	2.012s (~)	0.656s	15	1.02x
💻 Local	Next.js (Turbopack)	1.381s (-0.9%)	2.012s (~)	0.631s	15	1.04x
💻 Local	Express	1.418s (~)	2.007s (~)	0.588s	15	1.07x
🐘 Postgres	Next.js (Turbopack)	1.988s (-8.4% 🟢)	2.325s (-3.1%)	0.337s	13	1.50x
🐘 Postgres	Express	2.003s (-13.8% 🟢)	2.225s (-26.2% 🟢)	0.222s	14	1.51x
🌐 MongoDB	Next.js (Turbopack)	2.135s (~)	3.011s (~)	0.876s	10	1.61x
🌐 Turso	Next.js (Turbopack)	2.208s (~)	3.013s (~)	0.805s	10	1.66x
🐘 Postgres	Nitro	2.661s (+11.4% 🔺)	3.013s (~)	0.352s	10	2.00x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	2.801s (-25.1% 🟢)	3.588s (-22.2% 🟢)	0.787s	9	1.00x
▲ Vercel	Express	2.944s (-3.6%)	3.818s (-3.8%)	0.873s	8	1.05x
▲ Vercel	Next.js (Turbopack)	3.151s (+12.6% 🔺)	3.923s (-0.7%)	0.772s	8	1.13x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.all with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.981s (-8.7% 🟢)	2.159s (-29.1% 🟢)	0.178s	14	1.00x
💻 Local	Next.js (Turbopack)	2.095s (-3.8%)	2.945s (-5.8% 🟢)	0.850s	11	1.06x
💻 Local	Express	2.197s (-1.5%)	3.124s (-1.4%)	0.927s	10	1.11x
🌐 Redis	Next.js (Turbopack)	2.401s (-0.5%)	3.017s (~)	0.615s	10	1.21x
🌐 Starter	Next.js (Turbopack)	2.464s (+0.7%)	3.009s (~)	0.545s	10	1.24x
🐘 Postgres	Next.js (Turbopack)	2.640s (-4.0%)	3.017s (~)	0.378s	10	1.33x
🐘 Postgres	Express	2.890s (-7.1% 🟢)	3.011s (-20.3% 🟢)	0.122s	10	1.46x
🐘 Postgres	Nitro	2.996s (-10.3% 🟢)	3.361s (-16.6% 🟢)	0.365s	9	1.51x
🌐 Turso	Next.js (Turbopack)	4.685s (-1.3%)	5.179s (~)	0.494s	6	2.37x
🌐 MongoDB	Next.js (Turbopack)	4.787s (+0.6%)	5.179s (~)	0.392s	6	2.42x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Nitro	3.254s (-2.0%)	3.847s (-2.0%)	0.593s	8	1.00x
▲ Vercel	Express	3.540s (-6.3% 🟢)	4.122s (-13.6% 🟢)	0.582s	8	1.09x
▲ Vercel	Next.js (Turbopack)	3.618s (-4.7%)	4.338s (~)	0.720s	8	1.11x

🔍 Observability: Nitro | Express | Next.js (Turbopack)

Promise.race with 10 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	1.340s (-4.2%)	2.006s (~)	0.665s	15	1.00x
💻 Local	Next.js (Turbopack)	1.367s (-2.7%)	2.011s (~)	0.644s	15	1.02x
🌐 Starter	Next.js (Turbopack)	1.371s (+1.6%)	2.007s (~)	0.636s	15	1.02x
🌐 Redis	Next.js (Turbopack)	1.377s (-0.7%)	2.012s (~)	0.635s	15	1.03x
💻 Local	Express	1.414s (-1.3%)	2.005s (~)	0.592s	15	1.05x
🐘 Postgres	Express	1.642s (-9.9% 🟢)	2.010s (-6.7% 🟢)	0.368s	15	1.23x
🐘 Postgres	Next.js (Turbopack)	1.825s (-8.4% 🟢)	2.012s (-7.1% 🟢)	0.187s	15	1.36x
🐘 Postgres	Nitro	1.904s (+3.9%)	2.237s (+3.9%)	0.333s	14	1.42x
🌐 MongoDB	Next.js (Turbopack)	2.147s (~)	3.014s (~)	0.867s	10	1.60x
🌐 Turso	Next.js (Turbopack)	2.214s (-0.9%)	3.011s (~)	0.796s	10	1.65x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	2.627s (-13.8% 🟢)	3.680s (-6.5% 🟢)	1.053s	9	1.00x
▲ Vercel	Next.js (Turbopack)	2.740s (-1.5%)	3.778s (+5.5% 🔺)	1.038s	8	1.04x
▲ Vercel	Nitro	2.823s (-1.4%)	3.686s (-3.7%)	0.863s	9	1.07x

🔍 Observability: Express | Next.js (Turbopack) | Nitro

Promise.race with 25 concurrent steps

💻 Local Development

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	2.150s (-2.5%)	3.035s (-3.4%)	0.885s	10	1.00x
💻 Local	Next.js (Turbopack)	2.159s (-6.8% 🟢)	3.125s (-4.3%)	0.966s	10	1.00x
💻 Local	Express	2.269s (-3.4%)	3.198s (-2.3%)	0.929s	10	1.06x
🌐 Redis	Next.js (Turbopack)	2.375s (-1.7%)	3.013s (~)	0.638s	10	1.10x
🌐 Starter	Next.js (Turbopack)	2.454s (~)	3.010s (~)	0.556s	10	1.14x
🐘 Postgres	Express	2.621s (-1.2%)	3.023s (~)	0.402s	10	1.22x
🐘 Postgres	Next.js (Turbopack)	2.637s (+2.4%)	3.021s (~)	0.385s	10	1.23x
🐘 Postgres	Nitro	2.771s (~)	3.011s (-3.7%)	0.241s	10	1.29x
🌐 Turso	Next.js (Turbopack)	4.748s (+0.9%)	5.186s (~)	0.438s	6	2.21x
🌐 MongoDB	Next.js (Turbopack)	4.783s (~)	5.179s (~)	0.396s	6	2.22x

▲ Production (Vercel)

World	Framework	Workflow Time	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Express	3.099s (+0.9%)	3.810s (+3.2%)	0.711s	8	1.00x
▲ Vercel	Nitro	3.130s (+6.4% 🔺)	3.720s (+2.8%)	0.590s	9	1.01x
▲ Vercel	Next.js (Turbopack)	3.139s (-3.5%)	3.719s (~)	0.580s	9	1.01x

🔍 Observability: Express | Nitro | Next.js (Turbopack)

Stream Benchmarks(includes TTFB metrics)

workflow with stream

💻 Local Development

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
💻 Local	🥇 Nitro	0.117s (-32.0% 🟢)	0.999s (~)	0.014s (-13.9% 🟢)	1.020s (~)	0.903s	10	1.00x
🌐 Redis	Next.js (Turbopack)	0.121s (-9.6% 🟢)	1.005s (~)	0.000s (~)	1.017s (~)	0.896s	10	1.03x
🌐 Starter	Next.js (Turbopack)	0.126s (-3.5%)	1.005s (~)	0.000s (-100.0% 🟢)	1.011s (~)	0.884s	10	1.08x
💻 Local	Next.js (Turbopack)	0.138s (-1.4%)	1.003s (~)	0.017s (-1.2%)	1.028s (~)	0.890s	10	1.18x
💻 Local	Express	0.173s (-2.5%)	0.993s (~)	0.015s (+4.1%)	1.022s (~)	0.849s	10	1.47x
🌐 MongoDB	Next.js (Turbopack)	0.480s (~)	0.968s (~)	0.000s (-100.0% 🟢)	1.012s (~)	0.533s	10	4.09x
🌐 Turso	Next.js (Turbopack)	0.503s (+1.7%)	0.949s (-0.6%)	0.000s (-100.0% 🟢)	1.014s (~)	0.510s	10	4.29x
🐘 Postgres	Next.js (Turbopack)	1.212s (+12.5% 🔺)	1.831s (-7.2% 🟢)	0.000s (NaN%)	2.017s (~)	0.805s	10	10.33x
🐘 Postgres	Express	1.431s (-36.2% 🟢)	1.655s (-40.9% 🟢)	0.000s (-66.7% 🟢)	2.014s (-33.2% 🟢)	0.583s	10	12.20x
🐘 Postgres	Nitro	2.153s (-2.2%)	2.892s (+1.7%)	0.000s (NaN%)	3.016s (~)	0.863s	10	18.35x

▲ Production (Vercel)

World	Framework	Workflow Time	TTFB	Slurp	Wall Time	Overhead	Samples	vs Fastest
▲ Vercel	🥇 Next.js (Turbopack)	2.581s (-6.5% 🟢)	3.256s (-0.9%)	0.907s (+151.4% 🔺)	4.588s (+13.4% 🔺)	2.007s	10	1.00x
▲ Vercel	Express	2.700s (+4.6%)	3.253s (-3.2%)	0.934s (+100.7% 🔺)	4.707s (+10.2% 🔺)	2.007s	10	1.05x
▲ Vercel	Nitro	2.805s (+11.1% 🔺)	3.149s (+8.5% 🔺)	0.876s (+32.8% 🔺)	4.533s (+14.4% 🔺)	1.727s	10	1.09x

🔍 Observability: Next.js (Turbopack) | Express | Nitro

Summary

Fastest Framework by World

Winner determined by most benchmark wins

World	🥇 Fastest Framework	Wins
💻 Local	Nitro	8/8
🐘 Postgres	Next.js (Turbopack)	5/8
▲ Vercel	Express	4/8

Fastest World by Framework

Winner determined by most benchmark wins

Framework	🥇 Fastest World	Wins
Express	💻 Local	8/8
Next.js (Turbopack)	🌐 Redis	3/8
Nitro	💻 Local	8/8

Column Definitions

• Workflow Time: Runtime reported by workflow (completedAt - createdAt) - primary metric
• TTFB: Time to First Byte - time from workflow start until first stream byte received (stream benchmarks only)
• Slurp: Time from first byte to complete stream consumption (stream benchmarks only)
• Wall Time: Total testbench time (trigger workflow + poll for result)
• Overhead: Testbench overhead (Wall Time - Workflow Time)
• Samples: Number of benchmark iterations run
• vs Fastest: How much slower compared to the fastest configuration for this benchmark

Worlds:

• 💻 Local: In-memory filesystem world (local development)
• 🐘 Postgres: PostgreSQL database world (local development)
• ▲ Vercel: Vercel production/preview deployment
• 🌐 Starter: Community world (local development)
• 🌐 Turso: Community world (local development)
• 🌐 MongoDB: Community world (local development)
• 🌐 Redis: Community world (local development)
• 🌐 Jazz: Community world (local development)

---

📋 View full workflow run

[Copilot]

Pull request overview

This PR adds comprehensive AI SDK parity features to the DurableAgent class, significantly expanding its capabilities to match the AI SDK's streamText functionality while maintaining workflow durability guarantees.

Key Changes:

• Added generation settings support (temperature, maxOutputTokens, topP, topK, penalties, seed, etc.) at both constructor and per-stream levels
• Implemented step control via maxSteps (defaults to unlimited for backwards compatibility), toolChoice, and activeTools filtering
• Added lifecycle callbacks: onChunk, onStepFinish, onFinish, onError, and onAbort
• Implemented structured output parsing via experimental_output with Output.object()
• Added experimental context passing, telemetry, tool call repair, and stream transforms

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
packages/ai/src/index.ts	Exports DurableAgent, Output helper, and all new types for public API
packages/ai/src/agent/durable-agent.ts	Main implementation: adds type definitions, enhances DurableAgent class with generation settings, callbacks, structured output, and tool filtering
packages/ai/src/agent/stream-text-iterator.ts	Implements maxSteps limiting, abort signal handling, prepareStep enhancements for all generation settings, and callback integration
packages/ai/src/agent/do-stream-step.ts	Handles generation settings application to model calls, implements transforms, onChunk callback, and enables file/source/tool-input streaming
packages/ai/src/agent/durable-agent.test.ts	Comprehensive test coverage for all new features including generation settings, maxSteps, toolChoice, activeTools, callbacks, context, and repair functions
docs/content/docs/api-reference/workflow-ai/durable-agent.mdx	Extensive documentation with examples for all new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[pranaygp]

this doesn't work. You can't use setTimeout in the workflow context since it makes things non-deterministic. calling setTimeout will throw

[pranaygp]

I took out the example