Interview Patterns — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Interview Patterns (Agent Skill) and scored it 92/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 2 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
A systematic approach to gathering requirements before implementing features, reviewing code, or performing analysis. Using the AskUserQuestion tool ensures you understand the user's needs before taking action.
AskUserQuestion tool with:
- questions: Array of 1-4 questions
- Each question has:
- question: The full question text
- header: Short label (max 12 chars) for the chip/tag
- options: 2-4 choices with label and description
- multiSelect: true/false (allow multiple selections)Good Questions:
Bad Questions:
1. Receive user request
2. Identify what information is needed
3. Use AskUserQuestion with relevant questions
4. Wait for user responses
5. Summarize understanding back to user
6. Proceed with action based on answersUser: "Create a product detail screen"
Claude: [Uses AskUserQuestion]
- Question 1: What's the main purpose?
- Question 2: What components are needed?
- Question 3: Where does data come from?
User: [Selects answers]
Claude: "Got it! I'll create a product detail screen with:
- Image gallery, title, price, description
- Buy button with purchase flow
- Data from your REST API
[Shows ASCII preview]
[Then generates VIP+W code]"Look for:
1. **Screen Purpose**
- "What is the main purpose of this screen?"
- Options: Display data, Form input, Navigation hub, Settings
2. **Key Components**
- "What UI elements do you need?" (multiSelect)
- Options: Text labels, Text fields, Buttons, Images, Lists/Tables
3. **Data Source**
- "Where does the data come from?"
- Options: API/Network, Local database, User input, Static content
4. **Navigation**
- "How do users navigate from this screen?"
- Options: Push to detail, Present modal, Tab switching, Back only1. **Review Focus**
- "What should I focus on?"
- Options: Full review, Architecture only, Performance only, Security only
2. **Severity Threshold**
- "What issues should I flag?"
- Options: Blockers only, Blockers + warnings, Everything including minor
3. **Context**
- "What type of code is this?"
- Options: New feature, Bug fix, Refactoring, Legacy code1. **Scope**
- "What do you want to migrate?"
- Options: Single ViewController, Feature module, Entire app, Just analyze
2. **Current Pain Points**
- "What problems are you experiencing?" (multiSelect)
- Options: Bugs, Hard to test, Hard to change, Performance issues
3. **Testing Status**
- "Do you have existing tests?"
- Options: Yes - comprehensive, Yes - some, No tests, Not sure
4. **Timeline**
- "How urgent is this?"
- Options: Can take time, Medium urgency, Need it fast1. **Symptoms**
- "What issues are you seeing?" (multiSelect)
- Options: Slow scrolling, Slow launch, Memory warnings, UI freezes
2. **Target Area**
- "Where should I focus?"
- Options: Specific file/class, Entire feature, App-wide scan
3. **Success Metrics**
- "What's your target?"
- Options: 60fps scrolling, <2s launch, Reduce memory 50%, General improvement1. **Focus Area**
- "What's the primary concern?"
- Options: Data storage, Network security, Authentication, Full audit
2. **Sensitive Data**
- "What data does the app handle?" (multiSelect)
- Options: User credentials, Payment info, Personal health, Location
3. **Compliance**
- "Any compliance requirements?"
- Options: OWASP Mobile Top 10, HIPAA, PCI-DSS, General best practices// Based on user's interview answers:
if purpose == .displayData && dataSource == .api {
// Generate VIP+W with API Worker
} else if purpose == .formInput && dataSource == .userInput {
// Generate VIP+W with validation in Interactor
}
if reviewFocus == .securityOnly {
// Run security-focused review
} else if reviewFocus == .performanceOnly {
// Run performance-focused review
}If initial answers reveal ambiguity:
"You mentioned the data comes from multiple sources. Could you clarify:
- Which sources are primary vs. fallback?
- Should offline mode be supported?"| Agent | When to Interview |
|---|---|
| uikit-architect | Before creating any UI |
| legacy-migrator | Before analyzing or migrating |
| ios-code-reviewer | Before reviewing code |
| performance-optimizer | Before analyzing performance |
| memory-leak-detector | Before scanning for leaks |
| accessibility-auditor | Before auditing accessibility |
| ios-security-auditor | Before security audit |
| test-coverage-analyzer | Before analyzing coverage |
| appstore-reviewer | Before App Store review |
1. Interview → Gather requirements
2. Summarize → Confirm understanding
3. Preview → Show plan/ASCII/outline
4. Confirm → Get user approval
5. Execute → Perform the actionThe interview-first approach ensures:
Always use AskUserQuestion before significant actions unless the user has explicitly provided all needed information or requested to skip the interview.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.