---

description: "Build features guided by data insights, A/B testing, and continuous measurement" argument-hint: " [--experiment-type ab|multivariate|bandit] [--confidence 0.90|0.95|0.99]"

Data-Driven Feature Development Orchestrator

CRITICAL BEHAVIORAL RULES

You MUST follow these rules exactly. Violating any of them is a failure.

1. Execute steps in order. Do NOT skip ahead, reorder, or merge steps.
2. Write output files. Each step MUST produce its output file in .data-driven-feature/ before the next step begins. Read from prior step files — do NOT rely on context window memory.
3. Stop at checkpoints. When you reach a PHASE CHECKPOINT, you MUST stop and wait for explicit user approval before continuing. Use the AskUserQuestion tool with clear options.
4. Halt on failure. If any step fails (agent error, test failure, missing dependency), STOP immediately. Present the error and ask the user how to proceed. Do NOT silently continue.
5. Use only local agents. All subagent_type references use agents bundled with this plugin or general-purpose. No cross-plugin dependencies.
6. Never enter plan mode autonomously. Do NOT use EnterPlanMode. This command IS the plan — execute it.

Pre-flight Checks

Before starting, perform these checks:

1. Check for existing session

Check if .data-driven-feature/state.json exists:

• If it exists and status is "in_progress": Read it, display the current step, and ask the user:

  Found an in-progress data-driven feature session:
  Feature: [name from state]
  Current step: [step from state]
  
  1. Resume from where we left off
  2. Start fresh (archives existing session)
  

• If it exists and status is "complete": Ask whether to archive and start fresh.

2. Initialize state

Create .data-driven-feature/ directory and state.json:

{
  "feature": "$ARGUMENTS",
  "status": "in_progress",
  "experiment_type": "ab",
  "confidence_level": 0.95,
  "current_step": 1,
  "current_phase": 1,
  "completed_steps": [],
  "files_created": [],
  "started_at": "ISO_TIMESTAMP",
  "last_updated": "ISO_TIMESTAMP"
}

Parse $ARGUMENTS for --experiment-type and --confidence flags. Use defaults if not specified.

3. Parse feature description

Extract the feature description from $ARGUMENTS (everything before the flags). This is referenced as $FEATURE in prompts below.

---

Phase 1: Data Analysis & Hypothesis (Steps 1–3) — Interactive

Step 1: Exploratory Data Analysis

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Perform exploratory data analysis for $FEATURE"
  prompt: |
    You are a data scientist specializing in product analytics. Perform exploratory data analysis for feature: $FEATURE.

    ## Instructions
    1. Analyze existing user behavior data, identify patterns and opportunities
    2. Segment users by behavior and engagement patterns
    3. Calculate baseline metrics for key indicators
    4. Use modern analytics tools (Amplitude, Mixpanel, Segment) to understand current user journeys, conversion funnels, and engagement patterns
    5. Identify data quality issues or gaps that need addressing

    Provide an EDA report with user segments, behavioral patterns, and baseline metrics.

Save the agent's output to .data-driven-feature/01-eda-report.md.

Update state.json: set current_step to 2, add "01-eda-report.md" to files_created, add step 1 to completed_steps.

Step 2: Business Hypothesis Development

Read .data-driven-feature/01-eda-report.md to load EDA context.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Formulate business hypotheses for $FEATURE"
  prompt: |
    You are a business analyst specializing in data-driven product development. Formulate business hypotheses for feature: $FEATURE based on the data analysis below.

    ## EDA Findings
    [Insert full contents of .data-driven-feature/01-eda-report.md]

    ## Instructions
    1. Define clear success metrics and expected impact on key business KPIs
    2. Identify target user segments and minimum detectable effects
    3. Create measurable hypotheses using ICE or RICE prioritization frameworks
    4. Calculate expected ROI and business value

    Provide a hypothesis document with success metrics definition and expected ROI calculations.

Save the agent's output to .data-driven-feature/02-hypotheses.md.

Update state.json: set current_step to 3, add step 2 to completed_steps.

Step 3: Statistical Experiment Design

Read .data-driven-feature/02-hypotheses.md to load hypothesis context.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Design statistical experiment for $FEATURE"
  prompt: |
    You are a data scientist specializing in experimentation and statistical analysis. Design the statistical experiment for feature: $FEATURE.

    ## Business Hypotheses
    [Insert full contents of .data-driven-feature/02-hypotheses.md]

    ## Experiment Type: [from state.json]
    ## Confidence Level: [from state.json]

    ## Instructions
    1. Calculate required sample size for statistical power
    2. Define control and treatment groups with randomization strategy
    3. Plan for multiple testing corrections if needed
    4. Consider Bayesian A/B testing approaches for faster decision making
    5. Design for both primary and guardrail metrics
    6. Specify experiment runtime and stopping rules

    Provide an experiment design document with power analysis and statistical test plan.

Save the agent's output to .data-driven-feature/03-experiment-design.md.

Update state.json: set current_step to "checkpoint-1", add step 3 to completed_steps.

---

PHASE CHECKPOINT 1 — User Approval Required

You MUST stop here and present the analysis and experiment design for review.

Display a summary of the hypotheses from .data-driven-feature/02-hypotheses.md and experiment design from .data-driven-feature/03-experiment-design.md (key metrics, target segments, sample size, experiment type) and ask:

Data analysis and experiment design complete. Please review:
- .data-driven-feature/01-eda-report.md
- .data-driven-feature/02-hypotheses.md
- .data-driven-feature/03-experiment-design.md

1. Approve — proceed to architecture and implementation
2. Request changes — tell me what to adjust
3. Pause — save progress and stop here

Do NOT proceed to Phase 2 until the user selects option 1. If they select option 2, revise and re-checkpoint. If option 3, update state.json status and stop.

---

Phase 2: Architecture & Instrumentation (Steps 4–6)

Step 4: Feature Architecture Planning

Read .data-driven-feature/02-hypotheses.md and .data-driven-feature/03-experiment-design.md.

Use the Task tool:

Task:
  subagent_type: "data-engineering-backend-architect"
  description: "Design feature architecture for $FEATURE with A/B testing capability"
  prompt: |
    Design the feature architecture for: $FEATURE with A/B testing capability.

    ## Business Hypotheses
    [Insert contents of .data-driven-feature/02-hypotheses.md]

    ## Experiment Design
    [Insert contents of .data-driven-feature/03-experiment-design.md]

    ## Instructions
    1. Include feature flag integration (LaunchDarkly, Split.io, or Optimizely)
    2. Design gradual rollout strategy with circuit breakers for safety
    3. Ensure clean separation between control and treatment logic
    4. Support real-time configuration updates
    5. Design for proper data collection at each decision point

    Provide architecture diagrams, feature flag schema, and rollout strategy.

Save the agent's output to .data-driven-feature/04-architecture.md.

Update state.json: set current_step to 5, add step 4 to completed_steps.

Step 5: Analytics Instrumentation Design

Read .data-driven-feature/04-architecture.md.

Use the Task tool:

Task:
  subagent_type: "data-engineer"
  description: "Design analytics instrumentation for $FEATURE"
  prompt: |
    Design comprehensive analytics instrumentation for: $FEATURE.

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Experiment Design
    [Insert contents of .data-driven-feature/03-experiment-design.md]

    ## Instructions
    1. Define event schemas for user interactions with proper taxonomy
    2. Specify properties for segmentation and analysis
    3. Design funnel tracking and conversion events
    4. Plan cohort analysis capabilities
    5. Implement using modern SDKs (Segment, Amplitude, Mixpanel) with proper event taxonomy

    Provide an event tracking plan, analytics schema, and instrumentation guide.

Save the agent's output to .data-driven-feature/05-analytics-design.md.

Update state.json: set current_step to 6, add step 5 to completed_steps.

Step 6: Data Pipeline Architecture

Read .data-driven-feature/05-analytics-design.md.

Use the Task tool:

Task:
  subagent_type: "data-engineer"
  description: "Design data pipelines for $FEATURE"
  prompt: |
    Design data pipelines for feature: $FEATURE.

    ## Analytics Design
    [Insert contents of .data-driven-feature/05-analytics-design.md]

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Instructions
    1. Include real-time streaming for live metrics (Kafka, Kinesis)
    2. Design batch processing for detailed analysis
    3. Plan data warehouse integration (Snowflake, BigQuery)
    4. Include feature store for ML if applicable
    5. Ensure proper data governance and GDPR compliance
    6. Define data retention and archival policies

    Provide pipeline architecture, ETL/ELT specifications, and data flow diagrams.

Save the agent's output to .data-driven-feature/06-data-pipelines.md.

Update state.json: set current_step to "checkpoint-2", add step 6 to completed_steps.

---

PHASE CHECKPOINT 2 — User Approval Required

Display a summary of the architecture, analytics design, and data pipelines and ask:

Architecture and instrumentation design complete. Please review:
- .data-driven-feature/04-architecture.md
- .data-driven-feature/05-analytics-design.md
- .data-driven-feature/06-data-pipelines.md

1. Approve — proceed to implementation
2. Request changes — tell me what to adjust
3. Pause — save progress and stop here

Do NOT proceed to Phase 3 until the user approves.

---

Phase 3: Implementation (Steps 7–9)

Step 7: Backend Implementation

Read .data-driven-feature/04-architecture.md and .data-driven-feature/05-analytics-design.md.

Use the Task tool:

Task:
  subagent_type: "data-engineering-backend-architect"
  description: "Implement backend for $FEATURE with full instrumentation"
  prompt: |
    Implement the backend for feature: $FEATURE with full instrumentation.

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Analytics Design
    [Insert contents of .data-driven-feature/05-analytics-design.md]

    ## Instructions
    1. Include feature flag checks at decision points
    2. Implement comprehensive event tracking for all user actions
    3. Add performance metrics collection
    4. Implement error tracking and monitoring
    5. Add proper logging for experiment analysis
    6. Follow the project's existing code patterns and conventions

    Write all code files. Report what files were created/modified.

Save a summary to .data-driven-feature/07-backend.md.

Update state.json: set current_step to 8, add step 7 to completed_steps.

Step 8: Frontend Implementation

Read .data-driven-feature/04-architecture.md, .data-driven-feature/05-analytics-design.md, and .data-driven-feature/07-backend.md.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Implement frontend for $FEATURE with analytics tracking"
  prompt: |
    You are a frontend developer. Build the frontend for feature: $FEATURE with analytics tracking.

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Analytics Design
    [Insert contents of .data-driven-feature/05-analytics-design.md]

    ## Backend Implementation
    [Insert contents of .data-driven-feature/07-backend.md]

    ## Instructions
    1. Implement event tracking for all user interactions
    2. Build A/B test variants with proper variant assignment
    3. Add session recording integration if applicable
    4. Track performance metrics (Core Web Vitals)
    5. Add proper error boundaries
    6. Ensure consistent experience between control and treatment groups
    7. Follow the project's existing frontend patterns and conventions

    Write all code files. Report what files were created/modified.

Save a summary to .data-driven-feature/08-frontend.md.

Note: If the feature has no frontend component (pure backend/API/pipeline), skip this step — write a brief note in 08-frontend.md explaining why it was skipped, and continue.

Update state.json: set current_step to 9, add step 8 to completed_steps.

Step 9: ML Model Integration (if applicable)

Read .data-driven-feature/04-architecture.md and .data-driven-feature/06-data-pipelines.md.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Integrate ML models for $FEATURE"
  prompt: |
    You are an ML engineer. Integrate ML models for feature: $FEATURE if needed.

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Data Pipelines
    [Insert contents of .data-driven-feature/06-data-pipelines.md]

    ## Instructions
    1. Implement online inference with low latency
    2. Set up A/B testing between model versions
    3. Add model performance tracking and drift detection
    4. Implement automatic fallback mechanisms
    5. Set up model monitoring dashboards

    If no ML component is needed for this feature, explain why and skip.
    Write all code files. Report what files were created/modified.

Save a summary to .data-driven-feature/09-ml-integration.md.

Update state.json: set current_step to "checkpoint-3", add step 9 to completed_steps.

---

PHASE CHECKPOINT 3 — User Approval Required

Display a summary of the implementation and ask:

Implementation complete. Please review:
- .data-driven-feature/07-backend.md
- .data-driven-feature/08-frontend.md
- .data-driven-feature/09-ml-integration.md

1. Approve — proceed to validation and launch
2. Request changes — tell me what to fix
3. Pause — save progress and stop here

Do NOT proceed to Phase 4 until the user approves.

---

Phase 4: Validation & Launch (Steps 10–13)

Step 10: Analytics Validation

Read .data-driven-feature/05-analytics-design.md, .data-driven-feature/07-backend.md, and .data-driven-feature/08-frontend.md.

Use the Task tool:

Task:
  subagent_type: "data-engineer"
  description: "Validate analytics implementation for $FEATURE"
  prompt: |
    Validate the analytics implementation for: $FEATURE.

    ## Analytics Design
    [Insert contents of .data-driven-feature/05-analytics-design.md]

    ## Backend Implementation
    [Insert contents of .data-driven-feature/07-backend.md]

    ## Frontend Implementation
    [Insert contents of .data-driven-feature/08-frontend.md]

    ## Instructions
    1. Test all event tracking in staging environment
    2. Verify data quality and completeness
    3. Validate funnel definitions and conversion tracking
    4. Ensure proper user identification and session tracking
    5. Run end-to-end tests for data pipeline
    6. Check for tracking gaps or inconsistencies

    Provide a validation report with data quality metrics and tracking coverage analysis.

Save the agent's output to .data-driven-feature/10-analytics-validation.md.

Update state.json: set current_step to 11, add step 10 to completed_steps.

Step 11: Experiment Setup & Deployment

Read .data-driven-feature/03-experiment-design.md and .data-driven-feature/04-architecture.md.

Launch two agents in parallel using multiple Task tool calls in a single response:

11a. Experiment Infrastructure:

Task:
  subagent_type: "general-purpose"
  description: "Configure experiment infrastructure for $FEATURE"
  prompt: |
    You are a deployment engineer specializing in experimentation platforms. Configure experiment infrastructure for: $FEATURE.

    ## Experiment Design
    [Insert contents of .data-driven-feature/03-experiment-design.md]

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Instructions
    1. Set up feature flags with proper targeting rules
    2. Configure traffic allocation (start with 5-10%)
    3. Implement kill switches for safety
    4. Set up monitoring alerts for key metrics
    5. Test randomization and assignment logic
    6. Create rollback procedures

    Provide experiment configuration, monitoring dashboards, and rollout plan.

11b. Monitoring Setup:

Task:
  subagent_type: "general-purpose"
  description: "Set up monitoring for $FEATURE experiment"
  prompt: |
    You are an observability engineer. Set up comprehensive monitoring for: $FEATURE.

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Experiment Design
    [Insert contents of .data-driven-feature/03-experiment-design.md]

    ## Analytics Design
    [Insert contents of .data-driven-feature/05-analytics-design.md]

    ## Instructions
    1. Create real-time dashboards for experiment metrics
    2. Configure alerts for statistical significance milestones
    3. Monitor guardrail metrics for negative impacts
    4. Track system performance and error rates
    5. Define SLOs for the experiment period
    6. Use tools like Datadog, New Relic, or custom dashboards

    Provide monitoring dashboard configs, alert definitions, and SLO specifications.

After both complete, consolidate results into .data-driven-feature/11-experiment-setup.md:

# Experiment Setup: $FEATURE

## Experiment Infrastructure

[Summary from 11a — feature flags, traffic allocation, rollback plan]

## Monitoring Configuration

[Summary from 11b — dashboards, alerts, SLOs]

Update state.json: set current_step to 12, add step 11 to completed_steps.

Step 12: Gradual Rollout

Read .data-driven-feature/11-experiment-setup.md.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Create gradual rollout plan for $FEATURE"
  prompt: |
    You are a deployment engineer. Create a detailed gradual rollout plan for feature: $FEATURE.

    ## Experiment Setup
    [Insert contents of .data-driven-feature/11-experiment-setup.md]

    ## Instructions
    1. Define rollout stages: internal dogfooding → beta (1-5%) → gradual increase to target traffic
    2. Specify health checks and go/no-go criteria for each stage
    3. Define monitoring checkpoints and metrics thresholds
    4. Create automated rollback triggers for anomalies
    5. Document manual rollback procedures

    Provide a stage-by-stage rollout plan with decision criteria.

Save the agent's output to .data-driven-feature/12-rollout-plan.md.

Update state.json: set current_step to 13, add step 12 to completed_steps.

Step 13: Security Review

Read .data-driven-feature/04-architecture.md, .data-driven-feature/07-backend.md, and .data-driven-feature/08-frontend.md.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Security review of $FEATURE"
  prompt: |
    You are a security auditor. Perform a security review of this data-driven feature implementation.

    ## Architecture
    [Insert contents of .data-driven-feature/04-architecture.md]

    ## Backend Implementation
    [Insert contents of .data-driven-feature/07-backend.md]

    ## Frontend Implementation
    [Insert contents of .data-driven-feature/08-frontend.md]

    ## Instructions
    Review for: OWASP Top 10, data privacy and GDPR compliance, PII handling in analytics events,
    authentication/authorization flaws, input validation gaps, experiment manipulation risks,
    and any security anti-patterns.

    Provide findings with severity, location, and specific fix recommendations.

Save the agent's output to .data-driven-feature/13-security-review.md.

If there are Critical or High severity findings, address them now before proceeding. Apply fixes and re-validate.

Update state.json: set current_step to "checkpoint-4", add step 13 to completed_steps.

---

PHASE CHECKPOINT 4 — User Approval Required

Display a summary of validation and launch readiness and ask:

Validation and launch preparation complete. Please review:
- .data-driven-feature/10-analytics-validation.md
- .data-driven-feature/11-experiment-setup.md
- .data-driven-feature/12-rollout-plan.md
- .data-driven-feature/13-security-review.md

Security findings: [X critical, Y high, Z medium]

1. Approve — proceed to analysis planning
2. Request changes — tell me what to fix
3. Pause — save progress and stop here

Do NOT proceed to Phase 5 until the user approves.

---

Phase 5: Analysis & Decision (Steps 14–16)

Step 14: Statistical Analysis

Read .data-driven-feature/03-experiment-design.md and .data-driven-feature/02-hypotheses.md.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Create statistical analysis plan for $FEATURE experiment"
  prompt: |
    You are a data scientist specializing in experimentation. Create the statistical analysis plan for the A/B test results of: $FEATURE.

    ## Experiment Design
    [Insert contents of .data-driven-feature/03-experiment-design.md]

    ## Hypotheses
    [Insert contents of .data-driven-feature/02-hypotheses.md]

    ## Instructions
    1. Define statistical significance calculations with confidence intervals
    2. Plan segment-level effect analysis
    3. Specify secondary metrics impact analysis
    4. Use both frequentist and Bayesian approaches
    5. Account for multiple testing corrections
    6. Define stopping rules and decision criteria

    Provide an analysis plan with templates for results reporting.

Save the agent's output to .data-driven-feature/14-analysis-plan.md.

Update state.json: set current_step to 15, add step 14 to completed_steps.

Step 15: Business Impact Assessment Framework

Read .data-driven-feature/02-hypotheses.md and .data-driven-feature/14-analysis-plan.md.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Create business impact assessment framework for $FEATURE"
  prompt: |
    You are a business analyst. Create a business impact assessment framework for feature: $FEATURE.

    ## Hypotheses
    [Insert contents of .data-driven-feature/02-hypotheses.md]

    ## Analysis Plan
    [Insert contents of .data-driven-feature/14-analysis-plan.md]

    ## Instructions
    1. Define actual vs expected ROI calculation methodology
    2. Create a framework for analyzing impact on key business metrics
    3. Plan cost-benefit analysis including operational overhead
    4. Define criteria for full rollout, iteration, or rollback decisions
    5. Create templates for stakeholder reporting

    Provide a business impact framework and decision matrix.

Save the agent's output to .data-driven-feature/15-impact-framework.md.

Update state.json: set current_step to 16, add step 15 to completed_steps.

Step 16: Optimization Roadmap

Read .data-driven-feature/14-analysis-plan.md and .data-driven-feature/15-impact-framework.md.

Use the Task tool:

Task:
  subagent_type: "general-purpose"
  description: "Create post-launch optimization roadmap for $FEATURE"
  prompt: |
    You are a data scientist specializing in product optimization. Create a post-launch optimization roadmap for: $FEATURE.

    ## Analysis Plan
    [Insert contents of .data-driven-feature/14-analysis-plan.md]

    ## Impact Framework
    [Insert contents of .data-driven-feature/15-impact-framework.md]

    ## Instructions
    1. Define user behavior analysis methodology for treatment group
    2. Plan friction point identification in user journeys
    3. Suggest improvement hypotheses based on expected data patterns
    4. Plan follow-up experiments and iteration cycles
    5. Design cohort analysis for long-term impact assessment
    6. Create a continuous learning feedback loop

    Provide an optimization roadmap with follow-up experiment plans.

Save the agent's output to .data-driven-feature/16-optimization-roadmap.md.

Update state.json: set current_step to "complete", add step 16 to completed_steps.

---

Completion

Update state.json:

• Set status to "complete"
• Set last_updated to current timestamp

Present the final summary:

Data-driven feature development complete: $FEATURE

## Files Created
[List all .data-driven-feature/ output files]

## Development Summary
- EDA Report: .data-driven-feature/01-eda-report.md
- Hypotheses: .data-driven-feature/02-hypotheses.md
- Experiment Design: .data-driven-feature/03-experiment-design.md
- Architecture: .data-driven-feature/04-architecture.md
- Analytics Design: .data-driven-feature/05-analytics-design.md
- Data Pipelines: .data-driven-feature/06-data-pipelines.md
- Backend: .data-driven-feature/07-backend.md
- Frontend: .data-driven-feature/08-frontend.md
- ML Integration: .data-driven-feature/09-ml-integration.md
- Analytics Validation: .data-driven-feature/10-analytics-validation.md
- Experiment Setup: .data-driven-feature/11-experiment-setup.md
- Rollout Plan: .data-driven-feature/12-rollout-plan.md
- Security Review: .data-driven-feature/13-security-review.md
- Analysis Plan: .data-driven-feature/14-analysis-plan.md
- Impact Framework: .data-driven-feature/15-impact-framework.md
- Optimization Roadmap: .data-driven-feature/16-optimization-roadmap.md

## Next Steps
1. Review all generated artifacts and documentation
2. Execute the rollout plan in .data-driven-feature/12-rollout-plan.md
3. Monitor using the dashboards from .data-driven-feature/11-experiment-setup.md
4. Run analysis after experiment completes using .data-driven-feature/14-analysis-plan.md
5. Make go/no-go decision using .data-driven-feature/15-impact-framework.md