Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

AI SDK Handler · aiskillstore | DecimalAI

aiskillstore/ai-sdk-handler

SkillScore82Maintained

Integrate Vercel AI SDK for LLMs, Chatbots, Generative UI, and Agentic Workflows.

codingby aiskillstore

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

AI Engineer pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/aiskillstore-ai-sdk-handler/SKILL.md

decimalai skills pull aiskillstore-ai-sdk-handler --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/aiskillstore-ai-sdk-handler/SKILL.md https://app.decimal.ai/s/aiskillstore-ai-sdk-handler/SKILL.md

Raw file: /s/aiskillstore-ai-sdk-handler/SKILL.md ↗ · machine-readable summary at /s/aiskillstore-ai-sdk-handler.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 8d ago

Yes — +9% lift-9% tokens0% turns82% of whole cases pass with skill (18/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-06	✗→✓	▲ Improved	-15%	0%
case-10	✗→✓	▲ Improved	-21%	0%
case-17	✗→✓	▲ Improved	-27%	0%
case-20	✗→✓	▲ Improved	-5%	0%
case-08	✓→✗	▼ Worse	-52%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

aiskillstore / marketplace

skills/aayushbaniya2006/ai-sdk-handler/SKILL.md

View on GitHub ↗

Imported 8/3/2026Version v1

Skill contents

SKILL.md

AI SDK Handler

This skill defines how to implement Large Language Model (LLM) features using the Vercel AI SDK. It covers streaming chat, structured object generation, generative UI, and background agents.

Note: For Image/Video generation (Replicate, Fal.ai), continue to use ai-handler. Use ai-sdk-handler specifically for text, chat, and agentic text/JSON workflows.

When to Use

Chatbots: Building interactive chat interfaces (useChat, streamText).
Structured Data: Extracting JSON from text (generateObject).
Generative UI: Streaming React components directly from the server (streamUI).
Agents: Complex, multi-step reasoning tasks (often combined with Inngest).
Multimodal: Handling image inputs with text.

Capabilities

1. Streaming Chat

Tool: streamText (Server), useChat (Client).
Pattern: Create a route handler at src/app/api/chat/route.ts.
Auth: Wrap with withAuthRequired to protect the route.
UI: Use src/components/chat-ui/ for chat components.

2. Generative UI (RSC)

Tool: streamUI (Server).
Pattern: Return React components based on tool calls.
Use Case: Dashboards that build themselves, dynamic reports.

3. Structured Object Generation

Tool: generateObject.
Pattern: Define a Zod schema and get strictly typed JSON back.
Use Case: Populating database forms, extracting itinerary details, categorizing content.

4. Background Agents (with Inngest)

Tool: generateText / generateObject inside Inngest steps.
Why: Next.js Server Actions/Routes have timeouts (max 60s usually). Agents taking longer must run in the background.
Pattern:

Trigger Inngest event from UI.
Inngest function runs the LLM logic (loops, multi-step).
Store result in DB or notify user.
Docs: AI SDK Agents.

Best Practices

Streaming: Always prefer streaming for text generation > 2 seconds to improve perceived latency.
Auth: Never expose open AI routes. Always verify session.user.id.
Providers: Use @ai-sdk/openai or @ai-sdk/anthropic. Abstract the provider configuration in src/lib/ai/index.ts.
Backpressure: The AI SDK handles this automatically in streamText.
Caching: Use unstable_cache or KV stores if queries are repetitive.
Prompt Engineering: Keep prompts in a dedicated folder or constant file if they are complex.

Documentation & Examples

reference.md: Core setup and essential code snippets.
examples.md: Exhaustive examples covering:

Basic Chat
Generative UI
Structured Object Generation
Agents & Workflows (Loop Control)
Caching
Streaming Data
Reading UI Streams
Handling Backpressure
Multimodal Chat

Want to see if aiskillstore/ai-sdk-handler would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/22/2026, 00:01 UTC · 22 cases

Lift

+9%

82% vs 73%

Agent turns

0%

1.0 vs 1.0

Tokens

-9%

1,953.5 vs 2,153.05

fail → pass · 4pass → pass · 14fail → fail · 2pass → fail · 2

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I need to set up an interactive streaming chat endpoint in my Next.js App Router application. Can you show me how to implement a route handler that ensures user authentication and streams text responses back to the UI?	fail→fail	13,554	7,408	-45%	1	1	0%	2,718	2,145	-21%	0	0	—
case-02 I need to implement a text-to-image generation endpoint using Replicate in my Next.js App Router application. How should I set up the handler?	pass→pass	14,772	10,437	-29%	1	1	0%	3,013	2,913	-3%	0	0	—
case-03 How should I integrate a video generation workflow using Fal.ai into my web application server?	pass→pass	18,402	11,324	-38%	1	1	0%	3,146	2,851	-9%	0	0	—
case-04 How do I build a standard REST CRUD API endpoint in Next.js App Router to update user profile details in PostgreSQL?	pass→pass	17,027	11,039	-35%	1	1	0%	3,211	3,062	-5%	0	0	—
case-05 I want to extract structured travel itinerary details from user emails in Next.js and return strictly typed JSON. Which generation function and schema library should I use?	pass→pass	11,464	4,970	-57%	1	1	0%	1,990	1,661	-17%	0	0	—
case-06 I am building a multi-step reasoning agent in Next.js that runs for over 2 minutes and exceeds Server Action timeouts. What background queue tool should execute the agent?	fail→pass	14,729	6,414	-56%	1	1	0%	2,303	1,964	-15%	0	0	—
case-07 I want my Next.js server to stream React components directly to the client context based on tool calls. Which function should I call?	pass→pass	8,886	3,109	-65%	1	1	0%	1,562	1,305	-16%	0	0	—
case-08 In what project file path should central OpenAI and Anthropic provider configurations be abstracted for an App Router app?	pass→fail	11,308	1,512	-87%	1	1	0%	2,068	998	-52%	0	0	—
case-09 When securing AI API route handlers against unauthorized access, what specific session property must be verified?	pass→pass	9,840	1,735	-82%	1	1	0%	1,577	1,002	-36%	0	0	—
case-10 In which directory path should pre-built chat user interface components be placed?	fail→pass	6,871	1,486	-78%	1	1	0%	1,201	948	-21%	0	0	—
case-11 When an LLM text response is expected to take longer than 2 seconds, what execution pattern should be selected to minimize perceived latency?	pass→pass	12,804	6,042	-53%	1	1	0%	1,847	1,892	+2%	0	0	—
case-12 What Next.js caching strategy or storage mechanism should be applied to avoid re-running repetitive LLM queries?	pass→pass	14,568	10,736	-26%	1	1	0%	2,503	2,709	+8%	0	0	—
case-13 Where should complex system prompts and prompt templates be maintained in the project structure?	pass→pass	14,605	8,344	-43%	1	1	0%	2,481	2,224	-10%	0	0	—
case-14 Which React hook should be imported on the client to interact with a streaming text route endpoint?	pass→pass	4,476	2,480	-45%	1	1	0%	813	1,178	+45%	0	0	—
case-15 Which server function creates a text stream for interactive chat applications?	pass→pass	6,483	1,760	-73%	1	1	0%	999	1,064	+7%	0	0	—
case-16 Within background step functions executed by queue workers for agents, which text or object generation functions should be called?	pass→pass	16,866	6,926	-59%	1	1	0%	2,715	2,080	-23%	0	0	—
case-17 What function should be used to parse unstructured customer feedback directly into strongly typed database model payloads?	fail→pass	11,759	3,803	-68%	1	1	0%	1,934	1,411	-27%	0	0	—
case-18 How should a self-building dynamic analytics dashboard stream server-rendered React elements to the client?	pass→pass	19,782	14,049	-29%	1	1	0%	3,248	3,562	+10%	0	0	—
case-19 What middleware or route wrapper function should be applied to prevent unauthenticated access to AI endpoints?	fail→fail	12,019	2,311	-81%	1	1	0%	2,118	1,115	-47%	0	0	—
case-20 How should client backpressure be configured when streaming high-frequency text updates using streamText?	fail→pass	17,912	11,990	-33%	1	1	0%	2,920	2,788	-5%	0	0	—
case-21 What exact file path is recommended for the primary chat endpoint in Next.js App Router applications?	pass→fail	4,312	1,291	-70%	1	1	0%	777	984	+27%	0	0	—
case-22 How should image uploads accompanied by text prompts be submitted to an LLM endpoint?	pass→pass	13,295	12,446	-6%	1	1	0%	2,223	3,121	+40%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 72.7%
Pass rate with the skill: 81.8%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 2
Skill version measured: v1
Run completed: 8/22/2026, 00:01 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-06

Prompt

I am building a multi-step reasoning agent in Next.js that runs for over 2 minutes and exceeds Server Action timeouts. What background queue tool should execute the agent?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run82

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns-2% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 7 days ago

Other measured skills in the registry, with their headline benchmark lift.