hi-predict

---

name: hi-predict description: Five expert personas independently analyze proposed changes before implementation to catch architectural, security, performance, and UX issues early. Uses mind_mcp for project context and graph_mcp for code impact analysis. Produces GO/CAUTION/STOP verdict with consensus agreements, conflict resolutions, and risk mitigations. Use before major features, refactors, or risky changes. version: 1.0.0 last_updated: 2026-05-12 hooks: pre: - name: input-validation scope: [change_proposal, analysis_depth] enable_redaction: true phase: code_context: pre: [mcp-health-check] post: [progress-reporter] independent_analysis: post: [progress-reporter] consensus_debate: post: [progress-reporter] verdict: post: [progress-reporter] all_phases: post: [progress-reporter] post: - name: cleanup-handler paths: [predict-data/] keep: [*.json, *.md]

HI Predict

Five-expert-persona pre-analysis that debates proposed changes using MCP-assisted code context before any code is written.

When To Use

• Before implementing a major or high-risk feature
• Before a significant refactor or architecture change
• Evaluating competing technical approaches
• Stress-testing assumptions in a proposed design
• Gate check before committing to implementation

When NOT to Use

• Trivial or low-risk changes (single file, known pattern)
• Already-approved work with no open design questions
• Pure dependency upgrades with no API changes
• Documentation-only changes

Required Inputs

• Change proposal or feature description
• Optional: file paths to analyze (--files <glob>)
• Optional: specific concern areas to focus on
• Analysis depth: quick or deep

The 5 Personas

Persona	Focus	Core Questions
Architect	System design, scalability, coupling	Does this fit the architecture? Will it scale? What new coupling does it introduce?
Security	Attack surface, data protection, auth	What can be abused? Where is data exposed? Are auth boundaries respected?
Performance	Latency, memory, queries, resource usage	What is the latency impact? N+1 queries? Memory leaks? Resource contention?
UX	User experience, accessibility, error states	Is this intuitive? What does the error state look like? Accessible on all devices?
Devil's Advocate	Hidden assumptions, simpler alternatives	Why not do nothing? What is the simplest alternative? Which assumption could be wrong?

Input Validation & Security

proposal_validation:
  - proposal must be non-empty, 10-5000 characters
  - reject proposals containing only code (must be described in natural language)

depth_validation:
  allowed_values: ["quick", "deep"]
  default: "quick"

Performance & Operational Configuration

timeouts:
  phase_0_code_context_timeout: 180s
  phase_1_independent_analysis_timeout: 300s
  phase_2_consensus_debate_timeout: 180s
  phase_3_verdict_timeout: 60s

  total_workflow_timeout: 720s             # 12 minutes

progress_reporting:
  phase_start: "Phase {N} started: {phase_name}"
  persona_progress: "Persona {name} analyzing..."
  conflict_resolving: "Resolving conflict: {topic} ({architect} vs {security})"
  final_summary: "Verdict: {GO|CAUTION|STOP} — {agreement_count} agreements, {conflict_count} conflicts resolved"

---

Verdict Levels

Verdict	Meaning
GO	All personas aligned, no critical risks, proceed with confidence
CAUTION	Concerns exist but manageable — mitigations identified, proceed carefully
STOP	Critical unresolved issue found — redesign or more information needed

STOP Triggers (any one is sufficient)

• Security persona identifies auth bypass or data exposure with no viable mitigation
• Architect identifies fundamental design incompatibility requiring significant rework
• Performance persona identifies unacceptable latency or query explosion with no workaround
• Devil's Advocate exposes a false assumption that invalidates the entire approach

---

Orchestration Workflow

Phase 0: Code Context Gathering (3min)

steps:
  1. Parse change proposal and file scope
  2. Query mind_mcp for architecture docs and patterns
  3. Query graph_mcp for affected code areas and dependencies
  4. Build context package for persona analysis
  5. Report: "Phase 0 complete: Context gathered"

mcp_functions:
  - mind_mcp.hybrid_search [optional]
    params:
      query: "{proposal_keywords} architecture design"
      collection: "{collection}"
      limit: 10
    output:
      - results: relevant architecture docs
    expected: "Design patterns and architecture decisions"

  - graph_mcp.explore_graph [optional]
    params:
      query: "{affected_module_or_component}"
      limit: 50
    output:
      - nodes: affected components
      - edges: dependencies
    expected: "Code areas impacted by change"

  - graph_mcp.trace_flow [optional]
    params:
      start_node: "{entry_point_id}"
      max_depth: 5
    output:
      - flow: execution path
    expected: "Runtime flow through affected code"

Phase 1: Independent Analysis (5min)

steps:
  1. Each persona analyzes independently WITHOUT reading other personas' outputs
  2. Personas use MCP context + code reading for evidence
  3. Record findings per persona:
     - Architect: coupling, scalability, pattern fit
     - Security: attack surface, data exposure, auth gaps
     - Performance: latency, query patterns, resource usage
     - UX: user flow, error states, accessibility
     - Devil's Advocate: assumptions, alternatives, risks
  4. Report: "Phase 1 complete: {5} personas analyzed"

persona_output_format:
  architect:
    concerns: ["concern 1", "concern 2"]
    recommendations: ["action 1", "action 2"]
    confidence: "high|medium|low"

  security:
    threats: ["threat 1", "threat 2"]
    severity: "critical|high|medium|low"
    mitigations: ["mitigation 1"]

  performance:
    bottlenecks: ["bottleneck 1"]
    metrics_impact: "latency +Xms, queries +N"
    alternatives: ["alternative 1"]

  ux:
    issues: ["issue 1"]
    edge_cases: ["edge case 1"]
    a11y_concerns: ["a11y concern 1"]

  devils_advocate:
    assumptions_challenged: ["assumption 1"]
    simpler_alternatives: ["alternative 1"]
    worst_case: "description of worst case"

Phase 2: Consensus Debate (3min)

steps:
  1. Compare all persona outputs side-by-side
  2. Identify agreements — points where 4+ personas align
  3. Identify conflicts — points where personas meaningfully disagree
  4. For each conflict: weigh tradeoffs, determine resolution
  5. Document resolution with rationale
  6. Report: "Phase 2 complete: {agreements} agreements, {conflicts} conflicts resolved"

conflict_resolution_rules:
  - "Security vs Performance → Security wins unless performance makes system unusable"
  - "Architect vs UX → Defer to UX for user-facing features, Architect for backend"
  - "Devil's Advocate vs Everyone → If assumption is unvalidated, CAUTION"
  - "Any persona at Critical severity → cannot be GO"

Phase 3: Verdict & Report (1min)

steps:
  1. Synthesize all findings into consensus verdict
  2. Generate risk summary with severity and mitigation
  3. Produce actionable recommendations
  4. Format final report
  5. Report: "Phase 3 complete: Verdict={verdict}"

verdict_calculation:
  go: "0 Critical risks, <3 High risks, all mitigations clear"
  caution: "1-2 Critical mitigatable risks OR 3+ High risks"
  stop: "Any unmitigatable Critical risk OR false assumption detected"

---

Output Contract

# Prediction Report — {proposal_title}

**Date**: {timestamp}
**Depth**: {quick|deep}
**Verdict**: **GO** | **CAUTION** | **STOP**

## Executive Summary
{2-3 sentence summary of verdict and key reasoning}

## Agreements (all personas align)
- {Point 1}
- {Point 2}

## Conflicts & Resolutions

| Topic | Architect | Security | Performance | UX | Devil's Advocate | Resolution |
|-------|-----------|----------|-------------|-----|-----------------|------------|
| {topic} | {view} | {view} | {view} | {view} | {view} | {resolution} |

## Risk Summary

| Risk | Severity | Persona | Mitigation |
|------|----------|---------|------------|
| {risk} | Critical/High/Medium/Low | {persona} | {action} |

## Persona Detail

### Architect
- **Concerns**: {concerns}
- **Recommendations**: {recommendations}

### Security
- **Threats**: {threats}
- **Mitigations**: {mitigations}

### Performance
- **Bottlenecks**: {bottlenecks}
- **Impact**: {metrics}

### UX
- **Issues**: {issues}
- **Edge Cases**: {edge_cases}

### Devil's Advocate
- **Challenged Assumptions**: {assumptions}
- **Simpler Alternatives**: {alternatives}

## Recommendations
1. {Action} — {rationale}
2. {Action} — {rationale}

## Next Steps
- If GO: Proceed to implementation plan (use hi-plan)
- If CAUTION: Address mitigations before implementation
- If STOP: Redesign proposal and re-submit for prediction

---

Non-Negotiable Rules

• ✅ Personas MUST analyze independently — no cross-contamination during analysis phase
• ✅ Security persona findings always weighted higher for auth/data concerns
• ✅ Devil's Advocate must challenge at least one core assumption
• ✅ STOP verdict requires explicit documentation of what must change
• ✅ Every risk must have a concrete mitigation, not just identification
• ✅ Conflict resolutions must include rationale, not just the winner

Error Handling & Fallback Strategy

preflight:
  - check: "proposal_validation"
    action_on_failure: "abort_with_error"

fallback:
  when: "mcp_unavailable"
  steps:
    1. Skip MCP code context
    2. Analyze proposal text only
    3. Mark code-derived findings as lower confidence
    4. Note: "MCP unavailable — code context incomplete"

error_recovery:
  phase_1_independent_analysis:
    on_persona_timeout: "mark_persona_incomplete_and_continue"
  phase_2_consensus_debate:
    on_unresolvable_conflict: "document_as_unresolved_in_report"

Known Limitations

limitations:
  analysis_scope:
    - "Static analysis only — cannot predict runtime behavior"
    - "Analysis quality depends on proposal detail"
    - "MCP-unavailable mode produces less confident results"
  persona_limitations:
    - "Personas cannot ask clarifying questions — analysis is one-pass"
    - "Business logic nuances may be missed without domain expert input"

Deliverables

• prediction_report_{timestamp}.md — Full prediction report with all persona analyses, conflicts, verdict, and recommendations

References

• references/persona-playbook.md — Detailed analysis framework per persona with example prompts