New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Ma Design Interview

by michelangelo-ai in michelangelo-ai/michelangelo

Structured interview for designing and implementing changes to the Michelangelo platform.

Apache-2.0Auto-check passedAgent Workflows

Install Ma Design Interview

skills CLI
$ npx skills add michelangelo-ai/michelangelo --skill ma-design-interview -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install michelangelo-ai/michelangelo ma-design-interview --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/michelangelo-ai/michelangelo.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/ma-design-interview .claude/skills/ma-design-interview && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
ma-design-interview
GitHub stars
118
Token cost
~2.2k tokens
SKILL.md length
988 words
Files
2 (incl. references)
Skills in repo
9
Repo updated
First seen
Licence
Apache-2.0

At a glance

Structured interview for designing and implementing changes to the Michelangelo platform.

  • Works in 5 steps: Context Discovery → Interview → 5: Design Alternatives → …
  • The user says design interview
  • SKILL.md covers Interview rules, Phase 1: Context Discovery, Phase 2: Interview and Phase 2.5: Design Alternatives, plus 2 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Ma Design Interview is an agent skill from michelangelo-ai/michelangelo. Structured interview for designing and implementing changes to the Michelangelo platform. Auto-discovers proto schemas, API hooks, controllers, and existing configuration, then asks targeted questions to reach shared understanding before generating code. Domain-specific question trees live in references/ files. Use when the user says "design interview", "grill me", or when a task benefits from structured discovery before implementation.

Its SKILL.md is about 2.2k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files, including reference files (for example `references/ui-onboarding.md`).

It sits in Agent Workflows, covering Requirements gathering. The repository describes itself as: Michelangelo AI: Uber's end-to-end machine learning platform. The licence is Apache-2.0.

When your agent uses it

  • The user says design interview
  • A task benefits from structured discovery before implementation

Example prompts

  • “design interview”
  • “grill me”
  • “/ma-design-interview”

Workflow steps

5 steps, taken from the step headings in SKILL.md.

  1. Context Discovery
  2. Interview
  3. 5: Design Alternatives
  4. Generation
  5. Verification

What it can do on your machine

Read from SKILL.md and the folder at commit 491a9b2. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Links to these hosts (documentation or services it may open):

    • github.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Ma Design Interview loads about 2.2k tokens when it runs, and up to ~3.8k if it reads all its reference files. Until then it costs about 115 tokens; SKILL.md has 988 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~115
When it runs · the whole SKILL.md, loaded when a task matches
~2.2k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~3.8k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from michelangelo-ai/michelangelo at commit 491a9b2, republished under its Apache-2.0 licence (© michelangelo-ai). 988 words, ~2,170 tokens.

Download SKILL.mdSave it as .claude/skills/ma-design-interview/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
ma-design-interview
description
Structured interview for designing and implementing changes to the Michelangelo platform. Auto-discovers proto schemas, API hooks, controllers, and existing configuration, then asks targeted questions to reach shared understanding before generating code. Domain-specific question trees live in references/ files. Use when the user says "design interview", "grill me", or when a task benefits from structured discovery before implementation.

/ma-design-interview — Structured Design Interview

Interview a contributor about a design task — from backend entity onboarding to UI configuration to API design. Reads the codebase first, asks questions second, generates code at the end. Domain-specific guidance is loaded from references files.

Interview discipline derived from mattpocock/skills (MIT License).

Interview rules

  1. One question at a time. Never batch. Each answer informs the next question.
  2. Recommend an answer for every question. Propose what you'd do and why, then confirm.
  3. Finding facts is your job, not the user's. If the codebase can answer it, don't ask the contributor.
  4. Resolve dependencies in order. When decision A constrains decision B, settle A first.
  5. Stop only at shared understanding. Continue until every branch of the design tree has a resolved answer.

Phase 1: Context Discovery

Start by understanding what the contributor wants to do. The input can range from broad ("add new CRD to go services") to narrow ("add Y fields to the Z form").

Resolve resource names before discovery

Before launching any discovery, resolve every resource-like noun in the input to an actual proto file. Run ls proto/api/v2/*.proto and match the input against filenames.

Examples of why this matters:

  • "create a trigger run for a pipeline" → trigger_run.proto + pipeline.proto, NOT pipeline_run.proto. The input contains "run" and "pipeline" but the resource is TriggerRun, not PipelineRun.
  • "add notifications to pipeline runs" → pipeline_run.proto + notification.proto.

If the input maps cleanly to one or more proto files, proceed with those. If the mapping is ambiguous (e.g., multiple plausible resources, or no proto file matches), ask one scoping question to resolve.

Determine scope

Once the resource(s) are resolved:

  • New entity: resource name is sufficient (e.g., "Deployment"). Discover everything.
  • New surface on existing entity: resource name + which surface (list, detail, form).
  • Extend existing surface: resource name + which fields to add.
Read the architecture doc

Run find . -iname "ARCHITECTURE.md" -not -path "*/node_modules/*" -not -path "*/dist/*" to locate every architecture doc in the repo. More than one may exist (e.g. per-package under javascript/). If the command finds none, note that and move on.

For each result, skim its section headers, then read in full only the doc(s) covering the language area the task touches (go/, javascript/, python/); skip the rest.

1a. Discover schema artifacts

From the resource name (and any referenced messages), locate:

ArtifactConventionWhat to extract
Proto definitionproto/api/v2/<snake_case>.protoFields, types, enums, repeated fields, nested messages
API hooksgo/components/<lowercase>/apihook/BeforeCreate/BeforeUpdate validation — business rules beyond proto
Proto validationproto-go/api/v2/<snake_case>.pb.validation.goGenerated field constraints
Controllergo/components/<lowercase>/controller.goSide effects on create/update

If any path doesn't resolve, flag it — that's a directory structure problem, not something to work around.

1b. Discover existing UI config

Search packages/core/config/entities/ and app/config/ for existing configuration matching the resource name. Determine the scope:

  • Existing form found → extend mode. Read the form component, entity config, and TS type.
  • Existing list/detail config found → extend mode. Read the column/metadata configs.
  • Entity config exists but missing a surface → add the missing surface.
  • Nothing found → new entity. Will need PhaseEntityConfig, views, and form.

Tell the contributor what you found and confirm.

1c. Discover available UI components

Read these directories at runtime to build an inventory of what's already available:

DirectoryWhat it provides
javascript/packages/core/components/form/fields/Field components (string, select, boolean, date, etc.)
javascript/packages/core/components/form/layout/Layout components (FormGroup, FormRow, ArrayFormGroup, ArrayFormRow, FormGrid, etc.)
javascript/packages/core/components/form/validation/Built-in validators (required, min, max, minLength, maxLength, regex, url)
javascript/packages/core/components/cell/constants.tsCellType enum — available column/metadata renderers

Read the types.ts file for each component, not just the directory listing. The prop types determine what each component can do. Listing names without reading props leads to underusing what's already available or reinventing behavior that a prop already handles.

This inventory determines what can be built with existing components vs what needs custom work.

Show full SKILL.md (371 more words)Show less
1d. Build a field inventory

Scope depends on the task:

  • New entity or new surface: inventory all spec fields from the proto definition.
  • Extending an existing surface: inventory only the fields being added. If the new fields reference another proto message, inventory that message's fields instead.

For each field, capture: name, type, whether required, enum values (if applicable), and whether repeated. This table drives the interview.

Phase 2: Interview

Present what you discovered in Phase 1, then ask only about things the codebase can't answer.

Load the appropriate references file for the domain being discussed. The references file defines the question tree — what to ask, in what order, and what answers the codebase can provide vs what requires contributor input.

See references/ui-onboarding.md for UI configuration interviews.

Phase 2.5: Design Alternatives

When the interview surfaces a decision between meaningfully different approaches, don't ask the contributor to choose in the abstract. Build the alternatives and let them react to something concrete.

Trigger

Enter this phase when a Phase 2 question involves the shape of the solution — not a detail within a settled shape.

Triggers: "one thing or many?", "flat or nested?", "explicit or convention-based?", "user-configured or hardcoded?", "separate resources or inline?"

Does not trigger: "required or optional?", "what label?", "which renderer?"

Process
  1. Generate up to 3 alternatives, each labeled by its tradeoff posture — not "Option 1/2/3". The name should tell the contributor what they're optimizing for.
  2. Present them for comparison. For UI work with a running dev server, screenshot each one. For backend or API design, present the alternatives inline.
  3. The contributor picks one, or asks to revisit a specific alternative.
  4. No new interview questions during this phase. If the alternatives surface new questions, loop back to Phase 2.
Output

The chosen alternative becomes the input to Phase 3.

Phase 3: Generation

Generate the highest-quality implementation possible. The goal is a working prototype on a branch that either ships as-is (self-service) or gets refined by a frontend engineer.

Phase 4: Verification

Once generation is done, run /ma-sandbox-test-plan against the change to get real evidence it works — build/lint/test gates plus live integration scenarios in the sandbox — rather than declaring the interview complete on the strength of the generated code alone.

© michelangelo-ai, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file (references) in .claude/skills/ma-design-interview of michelangelo-ai/michelangelo.

  • SKILL.md
  • references/ui-onboarding.md

Open the folder on GitHubat commit 491a9b2

Compare with similar skills

Ma Design Interview next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Ma Design Interview compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Ma Design Interview this skillmichelangelo-ai/michelangelo118—~2.2kAutomated safety check: PassApache-2.0
Using Superpowersfarm-fe/farm5.6k35 repos~1.4kAutomated safety check: PassMIT
Interview Meaddyosmani/agent-skills104k6 repos~3.8kAutomated safety check: PassMIT
Brainstormingobra/superpowers297k1 repos~2.5kAutomated safety check: PassMIT
Grillingpietheinstrengholt/rssmonster56431 repos~510Automated safety check: PassMIT
Agentic Workflow Designerdotnet/Open-XML-SDK4.6k2 repos~3.5kAutomated safety check: PassMIT

Similar skills

  • Using Superpowers

    farm-fe/farm

    A skill your agent uses when starting any conversation - establishes how to find and use skills, requiring Skill tool invocation before ANY response including clarifying questions

    5.6k GitHub starsUsed in 35 repos~1.4k tokens
    Agent WorkflowsAuto-check passed

  • Interview Me

    addyosmani/agent-skills

    Asks one question at a time, each with a best guess attached, until the agent is about 95 percent sure what you really want, before any plan, spec or code.

    104k GitHub starsUsed in 6 repos~3.8k tokens
    Agent WorkflowsAuto-check passed

  • Brainstorming

    obra/superpowers

    Makes the agent clarify intent and agree on a design with you before writing any code, scaling the process from a quick spike to a written spec.

    297k GitHub starsUsed in 1 repo~2.5k tokens
    Agent WorkflowsAuto-check passed

  • Grilling

    pietheinstrengholt/rssmonster

    Grill the user relentlessly about a plan, decision, or idea.

    564 GitHub starsUsed in 31 repos~510 tokens
    Agent WorkflowsAuto-check passed

  • Agentic Workflow Designer

    dotnet/Open-XML-SDK

    Official

    Interviews you one question at a time about goal, trigger, permissions and data needs, then drafts a single agentic workflow markdown file.

    4.6k GitHub starsUsed in 2 repos~3.5k tokens
    Agent WorkflowsAuto-check passed

  • Ask User Question

    MemTensor/MemOS

    Shows a question as a modal in the interface to clarify a task, collect a preference or get approval, since the user cannot see terminal output.

    12k GitHub stars~1k tokensUpdated yesterday
    Agent WorkflowsAuto-check passed

More from michelangelo-ai/michelangelo

All 9 skills in this repo

  • Ma Sandbox Debug

    michelangelo-ai/michelangelo

    Tail logs, inspect pods, and diagnose unhealthy services in a running Michelangelo sandbox.

    118 GitHub stars~1k tokensUpdated today
    Auto-check passed

  • Ma Sandbox Deploy

    michelangelo-ai/michelangelo

    Build a Go service binary, package it into a Docker image, import into k3d, and deploy via helm sync.

    118 GitHub stars~1.4k tokensUpdated today
    Auto-check passed

  • Ma Sandbox Setup

    michelangelo-ai/michelangelo

    Canonical setup sequence for the Michelangelo local sandbox.

    118 GitHub stars~936 tokensUpdated today
    Auto-check passed

  • Ma Sandbox Test Plan

    michelangelo-ai/michelangelo

    Build, test, and verify a sandbox change across Go, JS, and Python.

    118 GitHub stars~2k tokensUpdated today
    Auto-check passed

  • Update Docs

    michelangelo-ai/michelangelo

    Update Michelangelo documentation. An agent skill from michelangelo-ai/michelangelo.

    118 GitHub stars~1.6k tokensUpdated today
    Auto-check passed

  • Ma Sandbox Reset

    michelangelo-ai/michelangelo

    Tear down the Michelangelo sandbox cluster and recreate it from scratch.

    118 GitHub stars~312 tokensUpdated today
    Auto-check passed

Categories

Questions about Ma Design Interview

What does Ma Design Interview do?

Structured interview for designing and implementing changes to the Michelangelo platform. Ma Design Interview is an agent skill from michelangelo-ai/michelangelo. Structured interview for designing and implementing changes to the Michelangelo platform.

When should I use Ma Design Interview?

Ma Design Interview fits situations like: the user says design interview; A task benefits from structured discovery before implementation.

How do I install Ma Design Interview in Claude Code?

Run `npx skills add michelangelo-ai/michelangelo --skill ma-design-interview -a claude-code`. Or copy the skill folder (.claude/skills/ma-design-interview in michelangelo-ai/michelangelo) into .claude/skills/ma-design-interview in your project. Claude Code loads it when a task matches its description.

How do I install Ma Design Interview in Codex?

Run `npx skills add michelangelo-ai/michelangelo --skill ma-design-interview -a codex`. Or copy the skill folder (.claude/skills/ma-design-interview in michelangelo-ai/michelangelo) into .agents/skills/ma-design-interview in your project. Codex loads it when a task matches its description.

Can I use Ma Design Interview in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add michelangelo-ai/michelangelo --skill ma-design-interview -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/ma-design-interview, .gemini/skills/ma-design-interview, .github/skills/ma-design-interview and .opencode/skills/ma-design-interview in your project.

What does Ma Design Interview need to run?

SKILL.md names no scripts, command-line tools or credentials: Ma Design Interview is instructions for the agent only.

Does Ma Design Interview access the network?

SKILL.md names 1 domain. As links in the text: github.com. This is read from the text; nothing was executed.

Is Ma Design Interview safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Ma Design Interview use?

Ma Design Interview is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Ma Design Interview use?

About 2.2k tokens (SKILL.md is roughly 8.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 1.7k tokens, read only when the agent opens those files.

What are the alternatives to Ma Design Interview?

Skills that share tags, products or a category with Ma Design Interview: Using Superpowers (farm-fe/farm, 5.6k stars), Interview Me (addyosmani/agent-skills, 104k stars), Brainstorming (obra/superpowers, 297k stars) and Grilling (pietheinstrengholt/rssmonster, 564 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Ma Design Interview?

michelangelo-ai (a GitHub organization) maintains it in michelangelo-ai/michelangelo, which has 118 GitHub stars. The repository holds 9 skills in this directory. The repository was last updated on October 10, 2026.

Source: michelangelo-ai/michelangelo on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.