# Phase 0: Validation Pipeline

> Take a raw idea and produce a data-backed GO / PIVOT / KILL recommendation before writing a single line of code.

---

## Pipeline Overview

```
Brain Dump (raw idea)
      |
      v
[0.1 Brain Dump Processor] -- Structure the idea
      |
      | Output: { problem, target_user, mvp_features, keywords, one_liner }
      v
[0.2 Market Research] -- Validate demand with real data
      |
      | Output: { demand_signals, trends, reddit_evidence, search_volume }
      v
[0.3 Competitor Analysis] -- Map the competitive landscape
      |
      | Output: { competitors[], gaps, differentiation, market_structure }
      v
[0.4 Validation Kill Check] -- Make the decision
      |
      | Output: GO / PIVOT / KILL + score + evidence + next_actions
      v
Phase 1 (if GO) | Pivot (if PIVOT) | Next idea (if KILL)
```

---

## Skills

| # | Skill | Purpose | MCP Servers Used |
|---|-------|---------|-----------------|
| 0.1 | [Brain Dump Processor](0.1-brain-dump/SKILL.md) | Structure raw ideas | None (pure reasoning) |
| 0.2 | [Market Research](0.2-market-research/SKILL.md) | Validate demand | Reddit Research, Google Trends, Exa.ai, Brave/Tavily |
| 0.3 | [Competitor Analysis](0.3-competitor-analysis/SKILL.md) | Map competition | Exa.ai, Firecrawl, Apify, Brave/Tavily |
| 0.4 | [Validation Kill Check](0.4-kill-check/SKILL.md) | GO/PIVOT/KILL decision | Consumes outputs from 0.1-0.3 |

---

## Required MCP Servers

**Minimum (free):**
- Reddit Research MCP (dialog-mcp) -- pain point discovery
- Google News+Trends MCP -- trend analysis
- Brave Search or Tavily -- web search

**Recommended (free tier):**
- Exa.ai MCP -- competitor discovery, deep search
- Firecrawl MCP -- structured data extraction
- Apify MCP -- review aggregation

**Optional (paid):**
- DataForSEO MCP -- actual search volume numbers ($50 deposit)

---

## How to Run

### Full Pipeline
```
Read my brain-dump.md and run the complete Phase 0 validation pipeline.
Use skills 0.1 through 0.4 in sequence.
```

### Individual Steps
```
Run skill 0.1 on my brain-dump.md
Run skill 0.2 using the output from 0.1
Run skill 0.3 using outputs from 0.1 and 0.2
Run skill 0.4 to make the GO/PIVOT/KILL decision
```

---

## Decision Gates

Each skill has a decision gate before proceeding:

| Gate | Pass Criteria | Fail Action |
|------|--------------|-------------|
| 0.1 -> 0.2 | Problem, target user, and keywords are specific | Ask user for clarification |
| 0.2 -> 0.3 | At least MODERATE demand signal from 1+ source | Flag risk, proceed anyway (competitor data may help) |
| 0.3 -> 0.4 | At least 3 competitors analyzed | If <3 found, document as risk (market may not exist) |
| 0.4 -> Phase 1 | Verdict is GO with score >= 7.0 | PIVOT: revisit with suggestions. KILL: stop. |

---

## Time Estimates

| Skill | Time (with MCPs) | Time (manual) |
|-------|-----------------|---------------|
| 0.1 Brain Dump | 5-15 min | 5-15 min |
| 0.2 Market Research | 15-30 min | 2-4 hours |
| 0.3 Competitor Analysis | 15-30 min | 3-6 hours |
| 0.4 Kill Check | 10-15 min | 30-60 min |
| **Total** | **45-90 min** | **6-12 hours** |

MCP servers reduce Phase 0 from a multi-day effort to under 2 hours.

---

*Part of the Thinking in Systems framework. Feeds into Phase 1-5 (Build).*