Mini Piv

Mini PIV Ralph - Lightweight Feature Builder

Arguments: $ARGUMENTS

Parse arguments:

FEATURE_NAME = $ARGUMENTS[0] or null (will ask user during discovery)
PROJECT_PATH = $ARGUMENTS[1] or current working directory

Examples:

• /mini-piv → Asks for feature name during discovery, uses cwd
• /mini-piv "add-user-auth" → Feature name provided, uses cwd
• /mini-piv "token-filters" /path/to/project → Both provided

Philosophy: Quick & Quality

"When you just want to build something without writing a PRD first."

Mini PIV Ralph is the lightweight sibling of the full PIV Ralph orchestrator. Same quality pipeline (Execute → Validate → Debug), but starts from a quick conversation instead of a PRD.

You are the orchestrator - stay lean at ~15% context, spawn fresh sub-agents for heavy lifting.

Required Reading by Role

DO NOT wing it. Sub-agents MUST read their instruction files before acting.

Visual Workflow

┌──────────────────────────────────────────────────────────┐
│              MINI PIV RALPH ORCHESTRATOR                    │
├──────────────────────────────────────────────────────────┤
│ 1. DISCOVERY (Orchestrator)                                │
│    Ask user 3-5 questions conversationally                 │
│    Extract: scope, touchpoints, deps, success, out-of-scope│
│                                                            │
│ 2. RESEARCH & PRP GENERATION (Fresh sub-agent)             │
│    ├─ Step 1: Codebase analysis (deep research)            │
│    ├─ Step 2: Generate PRP from discovery answers          │
│    └─ Output: PRPs/mini-{feature-name}.md                  │
│                                                            │
│ 3. EXECUTE (Executor sub-agent)                            │
│    → EXECUTION SUMMARY                                     │
│                                                            │
│ 4. VALIDATE (Validator sub-agent)                          │
│    → PASS / GAPS_FOUND / HUMAN_NEEDED                      │
│                                                            │
│ 5. DEBUG LOOP (if needed, max 3 iterations)                │
│    Spawn DEBUGGER → Re-validate → repeat or escalate       │
│                                                            │
│ 6. COMMIT (Orchestrator)                                   │
│    feat(mini): {description}                               │
└──────────────────────────────────────────────────────────┘

Step 1: Discovery Phase

1a. Determine Feature Name

If FEATURE_NAME is not provided:

1. Check recent conversation for context
2. If clear, propose a name and confirm
3. If unclear, ask directly

Normalize to kebab-case (e.g., "User Authentication" → "user-authentication").

1b. Check for Existing PRP

ls -la PROJECT_PATH/PRPs/ 2>/dev/null | grep -i "mini-{FEATURE_NAME}"

If exists, ask: "A PRP for '{FEATURE_NAME}' already exists. Overwrite, rename, or skip to execution?"

1c. Ask Discovery Questions (Conversational)

Present all questions in a single message. Adapt based on context.

Default question set:

I've got a few quick questions so I can build this right:

1. **What does this feature do?** Quick rundown — what's it supposed to accomplish?

2. **Where in the codebase does it live?** Specific files, folders, components?
   (Say "not sure" if you want me to figure it out)

3. **Any specific libraries, patterns, or existing code to follow?**

4. **What does "done" look like?** 1-3 concrete success criteria.

5. **Anything explicitly OUT of scope?** What should I NOT build right now?

Adapt by feature type:

• UI Components — also ask about styling, props, responsive behavior
• API Endpoints — also ask about request/response format, auth
• Smart Contracts — also ask about functions, events, security
• Integrations — also ask about external service details, error handling

Wait for user response, then extract structured answers.

1d. Structure Discovery Answers

feature:
  name: {FEATURE_NAME}
  scope: {Answer to Q1}
  touchpoints: {Answer to Q2}
  dependencies: {Answer to Q3}
  success_criteria: {Answer to Q4}
  out_of_scope: {Answer to Q5}

Step 2: Research & PRP Generation

CRITICAL: Spawn a FRESH sub-agent. Do NOT do this yourself.

Spawn a general-purpose sub-agent with this prompt:

MINI PIV RALPH: RESEARCH & PRP GENERATION
==========================================

You are generating a PRP for a lightweight, single-phase feature.

Project root: {PROJECT_PATH}
Feature name: {FEATURE_NAME}

## Discovery Input (from user)
{paste structured YAML from Step 1d}

## Step 1: Codebase Analysis
Run deep codebase analysis for this feature.
Save to: {PROJECT_PATH}/PRPs/planning/mini-{FEATURE_NAME}-analysis.md

## Step 2: Generate PRP (analysis context still loaded)
You already have the codebase analysis — use it directly.

### Discovery → PRP Translation Guide
| Discovery Answer | PRP Section |
|-----------------|-------------|
| Feature Scope (Q1) | Goal + What |
| Touch Points (Q2) | Implementation Blueprint task locations |
| Dependencies (Q3) | Context YAML, patterns, Known Gotchas |
| Success Criteria (Q4) | Success Criteria checklist + Final Validation |
| Out of Scope (Q5) | Explicit exclusions in What section |

### Template & Output
Use the PRP template: PRPs/templates/prp_base.md
Output PRP to: {PROJECT_PATH}/PRPs/mini-{FEATURE_NAME}.md

## Critical Rules
- Do BOTH steps yourself in sequence
- Follow the full PRP generation process
- DO NOT spawn sub-agents

Wait for completion.

Step 3: Spawn EXECUTOR Sub-Agent

Use the Task tool with subagent_type: "piv-executor":

EXECUTOR MISSION - Mini PIV Ralph
==================================

Execute the PRP at: {PROJECT_PATH}/PRPs/mini-{FEATURE_NAME}.md
Project root: {PROJECT_PATH}

Read references/execute-prp.md for the execution process.
Follow: Load PRP → Plan Thoroughly → Execute → Validate → Verify
Output EXECUTION SUMMARY with Status, Files, Tests, Issues.

Wait for completion.

Validation Sizing Decision

Before spawning a full validator, assess:

• <5 files changed, <100 lines, no external APIs → Quick validation (review changes yourself as orchestrator)
• Otherwise → Spawn full validator sub-agent (Step 4)

Step 4: Spawn VALIDATOR Sub-Agent

Use the Task tool with subagent_type: "piv-validator":

VALIDATOR MISSION - Mini PIV Ralph
===================================

Verify implementation of: {PROJECT_PATH}/PRPs/mini-{FEATURE_NAME}.md
Project root: {PROJECT_PATH}
Executor Summary: {paste EXECUTION SUMMARY}

Verify ALL requirements independently. Don't trust executor claims.
Output VERIFICATION REPORT with Grade, Checks, Gaps.

Process result:

• PASS → Step 6 (commit)
• GAPS_FOUND → Step 5 (debugger)
• HUMAN_NEEDED → Ask user

Step 5: Debug Loop (Max 3 iterations)

Spawn debugger using subagent_type: "piv-debugger":

DEBUGGER MISSION - Mini PIV - Iteration {I}
============================================

Project: {PROJECT_PATH}
PRP Path: {PROJECT_PATH}/PRPs/mini-{FEATURE_NAME}.md
Gaps: {GAPS}
Errors: {ERRORS}

Fix root causes, not symptoms. Run tests after each fix.
Output FIX REPORT with Status, Fixes Applied, Test Results.

After debugger: re-validate → PASS (commit) or loop (max 3) or escalate.

Step 6: Smart Commit

cd PROJECT_PATH && git status && git diff --stat

git add -A
git commit -m "feat(mini): implement {FEATURE_NAME}

- {bullet 1}
- {bullet 2}

Built via Mini PIV Ralph

Built with FTW (First Try Works) - https://github.com/SmokeAlot420/ftw"

Completion Summary

## MINI PIV RALPH COMPLETE

Feature: {FEATURE_NAME}
Project: {PROJECT_PATH}

### Artifacts
- PRP: PRPs/mini-{FEATURE_NAME}.md
- Analysis: PRPs/planning/mini-{FEATURE_NAME}-analysis.md

### Implementation
- Validation cycles: {N}
- Debug iterations: {M}

### Files Changed
{list}

### Commit
{hash}: {message}

All requirements verified and passing.

Error Handling

Executor Returns BLOCKED

Ask user: "Executor blocked on '{FEATURE_NAME}'. Issue: {description}. How should we proceed?"

Validator Returns HUMAN_NEEDED

Ask user: "Validator needs guidance. Question: {details}. Please advise."

3 Debug Cycles Exhausted

Ask user: "'{FEATURE_NAME}' failed validation after 3 fix attempts. Persistent issues: {list}. Need your guidance."

Sub-Agent Timeout/Failure

When a sub-agent times out or fails:

1. Check for partial work (files created, tests written)
2. Retry once with a simplified, shorter prompt
3. If retry fails, escalate to user with what was accomplished

Quick Reference

File Naming

PRPs/mini-{feature-name}.md                  # PRP file
PRPs/planning/mini-{feature-name}-analysis.md # Codebase analysis