Ideation-Claude

Multi-agent startup problem validator using Claude Code native sub-agents.

Quick Start

# 1. Clone the repo
git clone https://github.com/0xtechdean/ideation-claude.git
cd ideation-claude

# 2. Set up environment
cp .env.example .env
# Edit .env with your API keys (MEM0_API_KEY, SERPER_API_KEY, SLACK_BOT_TOKEN, SLACK_CHANNEL_ID)

# 3. Start Claude Code in the repo
claude

# 4. Ask Claude to validate a problem
> Validate the problem: "Legal research is too time-consuming for small law firms"

That's it! Claude reads CLAUDE.md, orchestrates 4 sub-agents, and delivers a comprehensive evaluation report.

Features

• Autonomous Execution: Uses ralph-wiggum plugin for hands-free operation
• Opus 4.5 Model: All agents run on Claude's most capable model
• Two-Phase Validation: Problem validation must pass before solution validation
• Early Elimination: Skip solution phase if problem_score < 6.0
• Parallel Agents: Problem validation agents run simultaneously
• Full Reports: Saved to reports/ and sent to Slack

Architecture

┌─────────────────────────────────────────────────────────────────┐
│                       Slack Trigger                             │
│    @Claude go to ideation-claude and validate "problem"         │
└─────────────────────────────┬───────────────────────────────────┘
                              │ triggers
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│                 ORCHESTRATOR (Claude Code)                      │
│                                                                 │
│   Phase 1: PROBLEM VALIDATION (parallel)                       │
│   ┌─────────────────────────────────────────────────────────┐  │
│   │  ┌──────────────┐  ┌──────────────┐                     │  │
│   │  │   market-    │  │  customer-   │                     │  │
│   │  │  researcher  │  │   solution   │                     │  │
│   │  └──────┬───────┘  └──────┬───────┘                     │  │
│   │         └────────┬────────┘                             │  │
│   │                  ▼                                      │  │
│   │         Calculate problem_score                         │  │
│   └─────────────────────────────────────────────────────────┘  │
│                              │                                  │
│           problem_score < 6.0? ──► ELIMINATE ──┐               │
│                              │                  │               │
│                              ▼ (if passes)      │               │
│   Phase 2: SOLUTION VALIDATION                  │               │
│   ┌──────────────────────────────────────────┐  │               │
│   │         ┌──────────────┐                 │  │               │
│   │         │ feasibility- │                 │  │               │
│   │         │    scorer    │                 │  │               │
│   │         └──────┬───────┘                 │  │               │
│   │                ▼                         │  │               │
│   │    combined = (problem×60%)+(solution×40%)  │               │
│   └──────────────────────────────────────────┘  │               │
│                              │                  │               │
│                              ◄──────────────────┘               │
│                              ▼                                  │
│   Phase 3: REPORT                                              │
│   ┌──────────────────────────────────────────────────────────┐ │
│   │              ┌──────────────────┐                        │ │
│   │              │   report-pivot   │ (+ pivots if failed)   │ │
│   │              └──────────────────┘                        │ │
│   └──────────────────────────────────────────────────────────┘ │
│                              │                                  │
│   Phase 4: Save report to reports/                             │
│   Phase 5: Send summary to Slack                               │
└─────────────────────────────────────────────────────────────────┘

How It Works

1. User in Slack: @Claude go to ideation-claude and validate "problem statement"
2. Claude Code reads CLAUDE.md and manages the entire flow
3. Phase 1: PROBLEM VALIDATION (parallel):
   • market-researcher - Market trends, TAM/SAM/SOM, market size score
   • customer-solution - Customer segments, pain points, WTP score
   • Calculate problem_score (severity, market, WTP, solution fit)
4. DECISION POINT: If problem_score < 6.0 → ELIMINATE → Skip to Phase 3
5. Phase 2: SOLUTION VALIDATION (only if problem passes):
   • feasibility-scorer - Competition, tech stack, resource requirements
   • Calculate solution_score and combined_score
6. Phase 3: report-pivot compiles final report (+ pivot suggestions if failed)
7. Phase 4: Saves report to reports/{problem}-evaluation-{session_id}.md
8. Phase 5: Sends summary to Slack

Native Sub-Agents

All agents are defined in .claude/agents/ as markdown files:

Agent	File	Role
Market Researcher	market-researcher.md	Market trends, pain points, TAM/SAM/SOM
Customer Solution	customer-solution.md	Customer segments, MVP design, pricing
Feasibility Scorer	feasibility-scorer.md	Competition, tech stack, 8-criteria scoring
Report & Pivot	report-pivot.md	Final report compilation, pivot suggestions

Evaluation Pipeline

PHASE 1: PROBLEM VALIDATION (parallel)
├── market-researcher  ─┐
└── customer-solution  ─┴──► Calculate problem_score
         │
         ▼
    problem_score < 6.0?
         │
    ┌────┴────┐
    │  YES    │──► ELIMINATE ──► report-pivot ──► Report + Pivot Suggestions
    └────┬────┘
         │ NO
         ▼
PHASE 2: SOLUTION VALIDATION
└── feasibility-scorer ──► Calculate solution_score
         │
         ▼
    combined = (problem × 60%) + (solution × 40%)
         │
         ▼
    combined >= 6.0? ──► PASS ──► report-pivot ──► Final Report
         │
    combined < 6.0?  ──► FAIL ──► report-pivot ──► Report + Pivot Suggestions

Scoring Criteria

Problem Validation (60% weight)

Criteria	Weight
Problem Severity	25%
Market Size	25%
Willingness to Pay	25%
Solution Fit	25%

Solution Validation (40% weight)

Criteria
Technical Viability
Competitive Advantage
Resource Requirements
Time to Market

Passing Threshold: Combined score >= 6.0/10

Usage

Option 1: Direct with Claude Code (Recommended)

# Clone and start
git clone https://github.com/0xtechdean/ideation-claude.git
cd ideation-claude
claude

# Then ask Claude:
> Validate the problem: "Your problem statement here"

# Or use ralph-wiggum for fully autonomous execution:
> /ralph-loop "Validate the problem: Your problem statement here" --max-iterations 30

Option 2: Via Slack

@Claude go to https://github.com/0xtechdean/ideation-claude and validate "Your problem statement here"

Performance

Metric	Time
Phase 1: Problem Validation	~4-5 min
Phase 2: Solution Validation	~3-4 min
Phase 3: Report Generation	~2-3 min
Full Flow (passes)	~10-12 min
Early Elimination	~5-7 min

Helper Scripts

Located in scripts/:

Script	Purpose
web_research.py	Web search via Serper API
mem0_helpers.py	Mem0 session management
analysis_tools.py	TAM/SAM calculations
slack_helpers.py	Slack notification formatting

Environment Variables

Variable	Required	Description
MEM0_API_KEY	Yes	Mem0 cloud storage
SERPER_API_KEY	Yes	Web search via Serper
SLACK_BOT_TOKEN	Yes	Slack Bot Token (xoxb-...)
SLACK_CHANNEL_ID	Yes	Channel for notifications

Project Structure

ideation-claude/
├── CLAUDE.md                    # Orchestrator instructions
├── .claude/
│   ├── settings.json            # Claude Code config (acceptEdits)
│   └── agents/
│       ├── market-researcher.md
│       ├── customer-solution.md
│       ├── feasibility-scorer.md
│       └── report-pivot.md
├── scripts/
│   ├── web_research.py
│   ├── mem0_helpers.py
│   ├── analysis_tools.py
│   └── slack_helpers.py
├── reports/                     # Generated evaluation reports
└── .env.example

License

MIT