# Ai901 Item Creator

> Generate AI-901 practice questions that feel like the real Microsoft Azure AI Fundamentals exam without copying it. Every item is grounded in current Microsoft Learn content, uses modern Microsoft product names (Microsoft Foundry, Azure AI Language, Azure AI Speech, Azure AI Vision, Azure Content Understanding), and follows the Microsoft Worldwide Learning Exam Writing Style Guide (WWL). Use when…

- **Type:** Skill
- **Install:** `agentstack add skill-timothywarner-org-ai901-cert-buddy-claude-ai901-item-creator`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [timothywarner-org](https://agentstack.voostack.com/s/timothywarner-org)
- **Installs:** 0
- **Category:** [Cloud & Infrastructure](https://agentstack.voostack.com/c/cloud-infrastructure)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [timothywarner-org](https://github.com/timothywarner-org)
- **Source:** https://github.com/timothywarner-org/ai901-cert-buddy-claude/tree/main/plugin/ai901-cert-buddy/skills/ai901-item-creator

## Install

```sh
agentstack add skill-timothywarner-org-ai901-cert-buddy-claude-ai901-item-creator
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Skill: ai901.practice_questions.exam_realistic

**Description:** Generate AI-901 practice questions that feel like the real Microsoft Azure AI Fundamentals exam without copying it. Every item is grounded in current Microsoft Learn content, uses modern Microsoft product names, and follows Microsoft-style exam item rules: scenario-first stems, plausible distractors, parallel choices, no trick wording.

## Exam facts (canonical)

- **Exam:** AI-901 -- Microsoft Azure AI Fundamentals (Microsoft Press video course companion)
- **Skills measured date:** April 15, 2026
- **Pass score:** 700
- **Study guide URL:** https://learn.microsoft.com/credentials/certifications/resources/study-guides/ai-901
- **Audience profile:** candidates at the beginning of their career in AI solution development, with conceptual knowledge of AI in Azure, Python coding syntax, and familiarity with Azure resources.

## Style precedence

The **Microsoft Worldwide Learning Exam Writing Style Guide** (WWL) is authoritative for every item. The **Microsoft Writing Style Guide** (MWSG) governs prose voice and tone where WWL is silent. When the two guides conflict, **WWL wins**.

Two MWSG conventions overridden by WWL for exam items:

- **No contractions** anywhere in an item.
- **All uppercase** for key names: TAB, ENTER, CTRL+ALT+DELETE.

## Grounding

**Required sources:**

- Microsoft Learn (truth source for objectives and product behavior). Access via the **Microsoft Learn MCP server** `ai901buddy-mslearn` using `microsoft_docs_search` and `microsoft_docs_fetch`.
- Microsoft Learn code samples (for Microsoft Foundry SDK, Azure AI Language, Azure AI Speech, Azure AI Vision, and Azure Content Understanding accuracy). Access via `microsoft_code_sample_search`.
- Canonical AI-901 objective domain file: `.claude/skills/_references/ai901-objective-domain.md` (verbatim mirror, dated 2026-05-05).

Grounding-first rule: query the Microsoft Learn MCP before any other source. Never substitute generic web search.

## AI-901 domains (canonical)

| Domain | Weight |
| --- | --- |
| Identify AI concepts and capabilities | 40-45% |
| Implement AI solutions by using Microsoft Foundry | 55-60% |

Pull specific objectives and sub-domains from `.claude/skills/_references/ai901-objective-domain.md`. If that file is unavailable, query the Microsoft Learn MCP server `ai901buddy-mslearn` for the AI-901 study guide.

## Style and word usage (WWL, mandatory)

### Question sentence

- Begin with **What** (stand-alone interrogative pronoun) or **Which** (followed by a noun).
- The auxiliary in question sentences is **should**. Do not use *can, must, might, do, would,* or *may* in question sentences.
- Approved stems: *What should you do?* / *What should you recommend?* / *Which service should you use?* / *Which two actions should you perform?* / *What are two possible ways to achieve this goal?*
- For Choose-N items, embed the count and use a plural noun: *Which two actions should you perform? Each correct answer presents part of the solution.*
- Avoid negatives. If a negative is unavoidable, **CAP** and **bold** it (for example, **NOT**).
- Goal statements use *You need to ...* or *You need to ensure that ...*
- Use *of the following* sparingly, only when answer choices are non-parallel.
- True/False stems are not allowed.

### Word usage

- **named**, not *called*, when introducing a name.
- **report**, not *complain*, when describing user feedback.
- Avoid **determine**. Use *verify, clarify, identify, set, discover, establish, calculate, decide*.
- Avoid **may**. Use *can* (valid action) or *might* (possibility).
- Avoid **would** in questions; use *should*. Avoid **would like**; use *want* or *need*.
- Avoid **using** alone; use *by using*. Avoid **with** in place of *by using*.
- Avoid specific determiners: *all, none, only, always, never*.
- Avoid indefinite qualifiers: *few, many, multiple, several, some, usually, a number of*.
- *site* is logical; *location* is physical -- never interchange.
- Use **plural**; do not use *(s)*.
- Present tense throughout.

### People (WWL legal rule)

- Never use a personal name in an item. Use **the user**.
- For multiple people, use **User1, User2, User3** (no spaces).

### Approved fictional companies (WWL Fictitious Names List)

Use only these companies. Always use the **entire** company name on every mention (write *Litware, Inc.*, not *Litware*; write *Contoso, Ltd.*, not *Contoso*). Randomize across the full list -- do not default to Contoso, Ltd.

| Company | Approved URL |
| --- | --- |
| A. Datum Corporation | adatum.com |
| Adventure Works Cycles | adventure-works.com |
| Alpine Ski House | alpineskihouse.com |
| Bellows College | bellowscollege.com |
| Best For You Organics Company | bestforyouorganics.com |
| Blue Yonder Airlines | blueyonderairlines.com |
| City Power & Light | cpandl.com |
| Coho Vineyard | cohovineyard.com |
| Coho Winery | cohowinery.com |
| Coho Vineyard & Winery | cohovineyardandwinery.com |
| Consolidated Messenger | consolidatedmessenger.com |
| Contoso, Ltd. | contoso.com |
| Contoso Pharmaceuticals | contoso.com |
| Contoso Suites | contososuites.com |
| Fabrikam, Inc. | fabrikam.com |
| Fabrikam Residences | fabrikamresidences.com |
| First Up Consultants | firstupconsultants.com |
| Fourth Coffee | fourthcoffee.com |
| Graphic Design Institute | graphicdesigninstitute.com |
| Humongous Insurance | humongousinsurance.com |
| Lamna Healthcare Company | lamnahealthcare.com |
| Liberty's Delightful Bakery & Cafe | libertysdelightfulbakeryandcafe.com |
| Litware, Inc. | litwareinc.com |
| Lucerne Publishing | lucernepublishing.com |
| Margie's Travel | margiestravel.com |
| Munson's Pickles and Preserves Farm | munsonspicklesandpreservesfarm.com |
| Nod Publishers | nodpublishers.com |
| Northwind Electric Cars | northwindelectriccars.com |
| Northwind Traders | northwindtraders.com |
| Proseware, Inc. | proseware.com |
| Relecloud | relecloud.com |
| School of Fine Art | fineartschool.net |
| Southridge Video | southridgevideo.com |
| Tailspin Toys | tailspintoys.com |
| The Phone Company | thephone-company.com |
| Trey Research | treyresearch.net |
| VanArsdel, Ltd. | vanarsdelltd.com |
| Wide World Importers | wideworldimporters.com |
| Wingtip Toys | wingtiptoys.com |
| Woodgrove Bank | woodgrovebank.com |

### Approved cities (WWL city list)

Use without state, province, or country: Atlanta, Boston, Chicago, Dallas, Denver, Detroit, Los Angeles, Mexico City, Montreal, New York, Ottawa, Quebec, San Diego, San Francisco, Seattle, Toronto, Vancouver; Amsterdam, Athens, Barcelona, Berlin, Brussels, Budapest, Copenhagen, Dublin, Frankfurt, Geneva, Glasgow, Hamburg, Helsinki, Lisbon, London, Madrid, Moscow, Oslo, Paris, Prague, Rome, Stockholm, Vienna, Warsaw; Beijing, Hong Kong, Kyoto, Osaka, Seoul, Shanghai, Taipei, Tokyo; Auckland, Bangkok, Calcutta, Jakarta, Manila, Melbourne, New Delhi, Perth, Singapore, Sydney; Ankara, Baghdad, Damascus, Riyadh, Tel Aviv; Cairo, Cape Town, Johannesburg, Lagos, Nairobi, Tangier; Bogota, Buenos Aires, Caracas, Lima, Rio de Janeiro, Santiago, Sao Paulo.

### Resource names (WWL conventions)

- **Servers:** Server1, Server2; Exch1, Exch2; SQL1, SQL2; DC1, DC2; DNS1, DNS2 (no space between the noun and the digit).
- **Computers:** Computer1, Computer2.
- **Applications:** App1, App2.
- **Subnets:** Subnet1, Subnet2. **Sites:** Site1, Site2.
- **Offices:** *main office*, *branch office*, *satellite office* (lowercase).
- Define each name on first mention.

### Answer choices (WWL)

- Default 4 choices (A-D) for AI-901 single-correct items: 1 correct + 3 distractors.
- Choose-N items: *N* correct + at least 2 distractors.
- Choices are **mutually exclusive** -- no overlap between any two.
- Choices are **parallel** in form, length, detail, and grammatical structure.
- Choices match the syntax of the question sentence (imperative stem -> imperative choices).
- All choices are sentence fragments (lowercase initial, no end punctuation) **OR** all complete sentences (capitalized, period). Do not mix.
- No *all of the above*, *none of the above*, *both A and B*.
- Eliminate redundant lead-in wording from every choice; move it to the stem.
- Distractors must reference real Microsoft services, Microsoft Foundry capabilities, Azure AI service features, or Azure resource kinds. Never invent fake names, models, or capabilities.
- No explanations inside answer choices.
- Order choices logically (numerical order; shortest to longest; related pairs together). Disable randomization when answers are numerical.

### Formatting

- **Bold** for UI element names and PowerShell cmdlets.
- **All uppercase** for key names: TAB, ENTER, CTRL+ALT+DELETE.
- File names labelled with the word *file*: *the deploy.bicep file*, not *deploy.bicep* alone.
- Sentence-style capitalization elsewhere, except product names and proper nouns.
- Oxford comma in every list of three or more.
- Plain ASCII only -- no curly quotes, en or em dashes. Use `--` and `->`.
- Spell out zero through nine; numerals for 10 and above.
- **should** = recommendation; **must** = requirement.
- Acronyms: spell out + acronym in parens on first mention (*natural language processing (NLP)*); acronym only after.

### Globalization

- Sentences 15-20 words.
- Active voice (subject + verb + object).
- Simple verb tenses.
- No `/` or `-` as punctuation. Replace *create/edit/display* with *create, edit, and display*.
- No possessives on product names.
- No nominalizations, noun stacks, idioms, slang, or US-only references.

## Terminology rename table (non-negotiable)

Always use current Microsoft product names. Never use a retired or legacy name, even if the user does. Silently map to the current name. If Microsoft Learn shows a different current name, prefer the Learn name.

| Retired / legacy name | Current name |
| --- | --- |
| Azure Active Directory (Azure AD) | Microsoft Entra ID |
| Azure AD tenant | Microsoft Entra tenant |
| Azure Cognitive Services (umbrella) | Azure AI services |
| Cognitive Services for Language | Azure AI Language |
| Cognitive Services for Speech | Azure AI Speech |
| Cognitive Services for Vision / Computer Vision | Azure AI Vision |
| Form Recognizer | Azure AI Document Intelligence |
| QnA Maker | Azure AI Language question answering |
| LUIS | Azure AI Language conversational language understanding |
| Azure OpenAI Service (standalone, post-rebrand) | Azure OpenAI in Microsoft Foundry |
| Azure AI Studio | Microsoft Foundry |
| Azure AI Foundry | Microsoft Foundry |
| Azure AI Foundry portal | Microsoft Foundry portal |
| Azure AI Foundry SDK | Microsoft Foundry SDK |
| Azure AI Foundry Agent Service | Microsoft Foundry agents |

## Guardrails

- **Exam integrity:** Do not recreate or paraphrase real exam questions. Do not reference braindumps. Write original scenarios and original stems every time.
- **Item quality:** Single skill measured per item. No trivia. No hidden requirements. One problem, one decision.
- **Audience fit:** Items target candidates at the beginning of their career in AI solution development. Assume conceptual knowledge of AI in Azure, Python coding syntax, and familiarity with Azure resources. Do not require advanced MLOps, data engineering, or production-architect depth.
- **Real features only:** Do not invent Microsoft products, services, models, or capabilities. Confine items to features described in the AI-901 study guide and validated through the Microsoft Learn MCP server `ai901buddy-mslearn`.

## Answer choice randomization (non-negotiable)

You MUST randomize which letter (A, B, C, or D) is the correct answer for each question. Do not default to any single letter position.

### Batch balance rule

When delivering 4 or more items in a single response:

- **4 to 7 items:** every letter (A, B, C, D) must be the correct answer at least once.
- **8 to 11 items:** every letter must appear at least twice.
- **12 or more items:** target an even split of 25 percent per letter; allowed drift is +/- 1.

**Procedure before delivery:**

1. Count how many times each letter is the correct answer across the drafted batch.
2. If any letter is missing or under the floor, pick one item where the correct answer falls in an over-represented letter and **rewrite the item** so that the correct answer moves to the under-represented letter. Re-shuffle distractor wording so the rewrite is still mutually exclusive and the rationale still maps cleanly.
3. Re-count. Repeat until the batch is balanced.
4. In the batch summary, state the final A/B/C/D distribution (for example, *Answer position distribution: A=2, B=3, C=2, D=3*).

Position must carry no signal.

## Workflow

1. Pull the target objective from `.claude/skills/_references/ai901-objective-domain.md` (or query the Microsoft Learn MCP server `ai901buddy-mslearn` for the AI-901 study guide if the file is unavailable).
2. Ground the intended correct behavior in Microsoft Learn using `microsoft_docs_search` first, then `microsoft_docs_fetch` for full-page detail.
3. If the item touches Microsoft Foundry SDK, Microsoft Foundry portal walkthroughs, Azure AI Language, Azure AI Speech, Azure AI Vision, or Azure Content Understanding configuration specifics, run `microsoft_code_sample_search` to confirm syntax.
4. Pick a random WWL-approved fictional company. Draft a workplace scenario stem that forces a real AI-901 candidate decision (which Azure AI service, which Microsoft Foundry capability, which model selection, which responsible AI consideration, which feature of Azure Content Understanding).
5. Randomly assign the correct answer to A, B, C, or D. Write 1 correct answer and 3 distractors based on common-but-wrong assumptions. After the batch is drafted, apply the **batch balance rule**.
6. Run a mutual exclusivity check on answer choices.
7. Run a **WWL style sweep**: stem starts with What/Which + should; no banned auxiliaries (can, must, might, do, would, may); no specific determiners; no indefinite qualifiers; no contractions; no parenthetical clauses; choices are parallel and consistent (all fragments or all complete sentences); resource names follow Server1/Computer1/App1 pattern; companies use the full WWL name; people use *the user* or User1/User2.
8. Run a terminology check: every product name matches the rename table above.
9. Run a candidate clarity check: single skill measured, no trivia, no hidden requirements, audience-appropriate depth.
10. Prepare rationale internally but do not deliver it yet (see delivery rules).

## Invalid answer handling

When presenting questions interactively:

- **hint:** Provide a clue that eliminates one distractor. Re-present the question with all four choices visible but the eliminated option noted.
- **skip** or **I do not know:** Immediately reveal the correct answer and full rationale (Phase 2), then move on.
- **Unrecognized input:** Prompt: *Please reply with **A**, **B**, **C**, or **D**. You can also enter **hint** for a clue or **skip** to see the answer.*

## Progress tracking

When multiple questions are requested:

- Prefix each with **Question N of M**.
- After the final question, present a summary: total correct, total incorrect, total skipped, and any weak domains identified.

## Scenario-first stem guidance

The stem opens with a workplace scenario before asking the question. Keep stems tight; one problem, one decision.

**Good example:**

> Tailspin Toys plans to add a customer-facing assistant that transcribes recorded support calls and identifies the language spoken in each segment. The solution must use a managed Azure AI service and must not require training a custom model. You need to recommend the service.
>
> What should you recommend?

**Bad example (no scenario, banned auxiliary):**

> Which Azure AI service can transcribe audio?

## Plausible distractor guidance

Distractors reference real Azure AI services, Microsoft Foundry capabilities, or mod

…

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [timothywarner-org](https://github.com/timothywarner-org)
- **Source:** [timothywarner-org/ai901-cert-buddy-claude](https://github.com/timothywarner-org/ai901-cert-buddy-claude)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-timothywarner-org-ai901-cert-buddy-claude-ai901-item-creator
- Seller: https://agentstack.voostack.com/s/timothywarner-org
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.