rad

Response-Aware Development (RAD) Skill

Systematic approach to identifying and verifying implicit assumptions in code using multi-model AI analysis. Prevents production failures by catching assumptions before they reach deployment.

When This Skill Activates

This skill automatically activates when user mentions:

• "assumption" or "assumptions"
• "verify assumptions" or "check assumptions"
• "list assumptions" or "find assumptions"
• "RAD" or "response-aware"
• "assumption tags" or "assumption verification"
• "critical assumptions" or "assumption analysis"

What This Skill Does

The RAD methodology identifies implicit assumptions that could cause production failures and routes them to appropriate AI models for verification based on risk level:

• #CRITICAL: Production-critical assumptions (payments, security, concurrency) → Premium models
• #ASSUME: Standard assumptions (state management, APIs, validation) → Dynamic free models
• #EDGE: Edge cases (browser compatibility, performance) → Fast free models

User Intent Detection and Routing

Assumption Verification

Triggers: "verify assumptions", "check my assumptions", "validate code assumptions", "run RAD verification"

Route to: /rad/verify workflow

Examples:

• "verify all assumptions in my code"
• "check critical assumptions only"
• "run assumption verification on changed files"

Assumption Listing

Triggers: "list assumptions", "find assumptions", "show all assumptions", "assumption inventory"

Route to: /rad/list workflow

Examples:

• "list all assumptions in the project"
• "show me critical assumptions"
• "find unverified assumptions"

RAD Testing

Triggers: "test RAD", "RAD demo", "test assumption system", "sample assumptions"

Route to: /rad/test workflow

Examples:

• "test the RAD system"
• "show me how RAD works"
• "create sample assumptions"

General RAD Questions

Triggers: "how does RAD work", "what is response-aware development", "assumption tagging guide"

Action: Load methodology context and explain RAD concepts

Examples:

• "explain the RAD methodology"
• "how do I tag assumptions?"
• "what are the assumption categories?"

Quick Command Reference

# Verify assumptions (tiered strategy, changed files only)
/rad/verify --strategy=tiered --scope=changed-files

# Verify only critical assumptions
/rad/verify --strategy=critical-only

# List all assumptions in project
/rad/list

# List assumptions in specific directory
/rad/list src/

# Test RAD system with samples
/rad/test

# Test with specific scenario
/rad/test advanced

RAD Methodology Overview

Assumption Tagging Standards

Tag assumptions during development to enable systematic verification:

// #CRITICAL: [category]: [assumption that could cause outages/data loss]
// #VERIFY: [defensive code required]
// Example: Payment processing, auth flows, concurrent writes

// #ASSUME: [category]: [assumption that could cause bugs]
// #VERIFY: [validation needed]
// Example: UI state, form validation, API responses

// #EDGE: [category]: [assumption about uncommon scenarios]
// #VERIFY: [optional improvement]
// Example: Browser compatibility, slow networks

Critical Assumption Categories (Mandatory Tagging)

• Timing Dependencies: State updates, async operations, race conditions
• External Resources: API availability, file existence, network connectivity
• Data Integrity: Type safety at boundaries, null/undefined handling
• Concurrency: Shared state, transaction isolation, deadlock potential
• Security: Authentication, authorization, input validation
• Payment/Financial: Transaction integrity, retry logic, rollback handling

Verification Workflow

1. Tag Assumptions: Claude tags during development, or manual tagging during review
2. Auto-Detection: Pre-commit hooks or explicit command trigger scan
3. Model Routing: Assumptions routed to appropriate models by risk level
4. Analysis: Fresh context prevents confirmation bias in verification
5. Fix Generation: Defensive code patterns generated automatically
6. Review & Apply: Developer reviews and applies fixes selectively

Integration with Other Skills

Pre-Commit Validation

Works seamlessly with quality skill:

/quality/precommit  # Includes RAD assumption verification

Security Auditing

Complements security skill:

/security/scan     # Static security analysis
/rad/verify        # Dynamic assumption verification

Agent Integration

For complex assumption analysis, this skill coordinates with specialized agents:

• Analysis Agent (mcp__zen-core__chat): Deep assumption pattern analysis
• Model Selector (mcp__zen-core__dynamic_model_selector): Optimal model selection
• Code Review Agent (via Task tool): Integration with code review workflows

Common Workflows

Before Committing Code

# 1. Quick check of changed files
/rad/verify --scope=changed-files

# 2. If critical assumptions found, verify with premium models
/rad/verify --strategy=critical-only --budget=premium

# 3. Review and apply fixes
git diff  # Review proposed changes
git add -p  # Selectively stage fixes

During Code Review

# 1. Get assumption inventory for PR
/rad/list

# 2. Verify all assumptions found
/rad/verify --strategy=tiered

# 3. Ensure no blocking issues
grep "❌ BLOCKING" assumption-report.md

Technical Debt Assessment

# 1. Full project scan
/rad/list .

# 2. Identify hotspots (files with high assumption density)
# 3. Prioritize verification by risk level
# 4. Schedule verification sprints

Success Metrics

• Detection Rate: Assumptions tagged per 1000 lines of code
• Fix Rate: Percentage of assumptions requiring defensive code
• Cost Efficiency: >90% of verifications using free models
• Production Impact: Reduction in assumption-related incidents

Resources

• Complete Methodology: See context/methodology.md
• Tagging Standards: See context/tagging-standards.md
• Verification Workflow: See workflows/verify.md
• Global Standards: See CLAUDE.md > Response-Aware Development

Consolidated from verify-assumptions-smart, list-assumptions, and test-rad commands.