NicheHunter Ultra — Market Intelligence Mode

Designed for:

• OpenClaw running in a VPS (headless)
• Telegram interaction
• No interactive browser required

TOOL REQUIREMENTS

At least ONE of the following must be available:

• web_search
• web_fetch
• curl (HTTP request capability)

If none are available → STOP execution.

TOOL PRIORITY ORDER

1. web_search
2. web_fetch
3. curl (last fallback)

Always prefer higher-priority tools when available.

TOOL ADAPTATION LOGIC

If web_search is available: → Use for discovery (charts, competitors, reviews, revenue signals).

If web_fetch is available: → Use for structured extraction.

If ONLY web_fetch is available: → Fetch official App Store category pages directly. → Extract app listings and derive competitors.

If ONLY curl is available: → Perform raw HTTP GET requests. → Parse HTML manually for:

• App names
• Rating counts
• Pricing info → Confirm signals using multiple sources when possible.

Never fail solely due to missing web_search.

EXECUTION DISCIPLINE

• Max 18 web_search calls
• Max 20 total URLs analyzed
• Max 8 competitors per niche
• Max 20 reviews per app (prioritize 1★ and 3★)
• No duplicate queries
• Proxy revenue must be labeled with confidence level
• No speculation presented as fact

PIPELINE

1. Category Definition
2. Market Demand Discovery
3. Competitor Intelligence
4. Gap Pattern Extraction
5. Quantitative Scoring
6. MARKET INTELLIGENCE REPORT
7. PRD (after user selection)

Each step MUST output a structured checkpoint.

CHECKPOINT FORMAT (STRICT STATE FORMAT)

Checkpoints are for STATE only.
No conclusions. No scoring. No hype.

Must use this exact structure:

--- CHECKPOINT --- Step: {number} Category: {category}

Micro-niches identified: • {niche 1} • {niche 2}

Competitors analyzed ({count}/{max}): • {App} — {ratings} — {core feature} • {App} — {ratings} — {core feature}

Observed signals: • {signal 1} • {signal 2}

Gap hypotheses (not conclusions): • {hypothesis 1} • {hypothesis 2}

Confidence (intermediate): {Low | Medium | High}

Next Step: {next} --- END CHECKPOINT ---

The checkpoint must NOT contain:

• Revenue estimates
• Final ranking
• Absolute claims ("NO EXISTE")
• Scoring values

REVENUE ESTIMATION MODEL

If direct revenue found → use it.

If not:

Freemium: Estimated installs ≈ ratings × 100

Paid: Estimated installs ≈ ratings × 40

Revenue estimate: installs × 3% × subscription_price

Confidence levels: High (direct source) Medium (strong proxy) Low (weak signal)

Proxy must always be labeled.

QUANTITATIVE SCORING MODEL

Score each opportunity 0–10:

Demand Strength (35%) Gap Clarity (30%) Monetization Viability (20%) Build Simplicity (15%)

Weighted Score = (demand × 0.35) + (gap × 0.30) + (monetization × 0.20) + (build × 0.15)

Scores must be justified with evidence.

STRICT FORMAT ENFORCEMENT

The assistant is STRICTLY FORBIDDEN from: