anysite-competitor-analyzer

What this skill does

# Competitor Analyzer

Systematic framework for gathering and analyzing competitive intelligence using Anysite MCP v2 tools.

## Tool Interface (v2)

All data fetching uses the unified v2 meta-tools:

- **`execute(source, category, endpoint, params)`** - Fetch data. Returns first 10 items + `cache_key`. If `next_offset` is present, use `get_page()` to load more.
- **`get_page(cache_key, offset, limit)`** - Paginate through cached results from a previous `execute()`. Data cached 7 days.
- **`query_cache(cache_key, conditions, sort_by, aggregate, group_by)`** - Filter, sort, count, or aggregate already-fetched data without new API calls.
- **`export_data(cache_key, format)`** - Export full dataset as CSV, JSON, or JSONL. Returns a download URL.
- **`discover(source, category)`** - Inspect available endpoints and params before calling `execute()`.

**Error handling:** If `execute()` returns an error with `llm_hint`, follow the hint to fix the call. Common issues: wrong URN format, alias not found (search first), fsd_company vs company: prefix mismatch.

## When to Use This Skill

Trigger this skill when users ask to:
- "Analyze [competitor name]"
- "Research our competitors"
- "Create a competitive analysis of [company]"
- "How does [competitor] position themselves?"
- "What are [competitor]'s strengths and weaknesses?"
- "Compare our product with [competitor]"
- "Who are our main competitors?"
- "Build a battle card for [competitor]"

## Quick Start

**For single competitor analysis:**
```bash
# 1. Generate analysis template
python scripts/analyze_competitor.py "Competitor Name" "https://competitor.com"

# 2. Use Anysite v2 tools to gather data (see workflow below)

# 3. Fill in the JSON template with findings

# 4. Generate final report
python scripts/analyze_competitor.py "Competitor Name" "https://competitor.com" | \
    python -c "import sys,json; exec('from scripts.analyze_competitor import format_markdown_report; print(format_markdown_report(json.load(sys.stdin)))')" \
    > /mnt/user-data/outputs/competitor_report.md
```

## Analysis Workflow

### Phase 1: Foundation (15-20 min)

**Step 1: Initialize Analysis Structure**

Run the analysis script to create structured template:
```bash
python scripts/analyze_competitor.py "Competitor Name" "https://competitor.com" > /tmp/analysis.json
```

**Step 2: Web Presence Reconnaissance**

Scrape key pages to understand positioning:
```python
# Homepage - core messaging
execute("webparser", "parse", "parse", {
    "url": "https://competitor.com",
    "only_main_content": true,
    "strip_all_tags": true
})

# Pricing - cost structure
execute("webparser", "parse", "parse", {
    "url": "https://competitor.com/pricing",
    "only_main_content": true
})

# About - company background
execute("webparser", "parse", "parse", {
    "url": "https://competitor.com/about",
    "only_main_content": true,
    "extract_contacts": true
})
```

**Extract from homepage:**
- H1/H2 headlines → positioning_statement
- Feature bullets → core_features
- Customer logos → customer_logos
- Value prop → value_proposition

**Extract from pricing:**
- Tier names and prices → pricing.tiers
- Cost per unit → pricing.unit_economics
- Free tier details → pricing.free_tier_limits
- Entry price → pricing.entry_price

**Extract from about:**
- Company description → company_overview.description
- Location → company_overview.headquarters
- Team size hints → company_overview.employee_count

### Phase 2: LinkedIn Intelligence (10-15 min)

**Step 3: Find Company Profile**

```python
# Search for company
execute("linkedin", "search", "search_companies", {
    "keywords": "competitor name",
    "count": 5
})

# Get detailed profile using slug from search results
execute("linkedin", "company", "company", {
    "company": "company-slug-from-search"
})
```

**Extract:**
- `follower_count` → Online presence indicator
- `employee_count` → Company size
- `description` → Self-positioning
- `headquarters` → Location
- `specialties` → Keywords they emphasize

**Step 4: Analyze Team & Growth**

```python
# Check employee growth signals (use search_users with current_company filter)
execute("linkedin", "search", "search_users", {
    "current_company": "company-slug",
    "keywords": "engineer developer",
    "count": 50
})

# Find leadership
execute("linkedin", "search", "search_users", {
    "current_company": "company-slug",
    "title": "CEO founder",
    "count": 10
})

# Get employee stats breakdown (functions, seniority, growth trends)
# First get company URN from company profile, convert fsd_company to company: prefix
execute("linkedin", "company", "company_employee_stats", {
    "urn": {"type": "company", "value": "COMPANY_ID_FROM_URN"}
})
```

**Use findings to assess:**
- Team size → growth_indicators.employee_growth
- Eng:sales ratio → GTM strategy signal
- Recent hires → growth phase indicator

**Tip:** Use `query_cache()` on the employee search results to filter by title or sort by relevance without re-fetching:
```python
query_cache(cache_key, conditions=[{"field": "headline", "operator": "contains", "value": "engineer"}])
```

**Step 5: Content Strategy**

```python
# Analyze posting activity
# Use company: prefix URN from the company profile (convert fsd_company:{id} to company:{id})
execute("linkedin", "company", "company_posts", {
    "urn": {"type": "company", "value": "COMPANY_ID_FROM_URN"},
    "count": 20
})
```

**Analyze posts for:**
- Frequency → content_strategy.blog_frequency
- Themes → content_strategy.key_topics
- Engagement → online_presence.linkedin.engagement_quality
- Tone → content_strategy.tone_of_voice

**Tip:** Use `query_cache()` to aggregate engagement metrics across fetched posts:
```python
query_cache(cache_key, aggregate=[{"field": "comment_count", "function": "avg"}])
```

### Phase 3: Deep Social & Community Research (20-30 min)

**Step 6: Twitter Deep Dive**

**A. Company Account Analysis**
```python
# Get profile stats
execute("twitter", "user", "get", {
    "username": "competitor_handle"
})

# Recent activity (analyze more posts)
execute("twitter", "user_tweets", "get", {
    "username": "competitor_handle",
    "count": 100
})
# If next_offset returned, use get_page() to load more:
# get_page(cache_key, offset=next_offset, limit=50)
```

**Extract from company account:**
- Followers → reach indicator
- Tweet frequency → activity level
- Content mix (product updates, thought leadership, customer engagement)
- Response time to mentions
- Tone of voice
- Most engaging tweets (viral content patterns)

**Tip:** Use `query_cache()` to find top-performing tweets:
```python
query_cache(cache_key, sort_by=[{"field": "favorite_count", "order": "desc"}])
```

**B. Founder/Executive Twitter Presence**
```python
# Find and analyze founder accounts
execute("twitter", "user", "get", {
    "username": "founder_handle"
})

execute("twitter", "user_tweets", "get", {
    "username": "founder_handle",
    "count": 100
})
```

**Leadership Twitter signals:**
- Personal brand strength
- Technical credibility (what they share)
- Customer engagement quality
- Industry thought leadership
- Follower quality (who follows them)
- Retweet patterns (what they amplify)

**C. Brand Mentions & Sentiment**
```python
# Comprehensive mention search
execute("twitter", "search", "search_posts", {
    "query": "competitor_name OR @handle OR #competitor_hashtag",
    "count": 200
})

# Problem/complaint mentions
execute("twitter", "search", "search_posts", {
    "query": "competitor_name (problem OR issue OR bug OR slow OR expensive)",
    "count": 100
})

# Positive sentiment
execute("twitter", "search", "search_posts", {
    "query": "competitor_name (love OR great OR amazing OR best OR solved)",
    "count": 100
})

# Competitive mentions
execute("twitter", "search", "search_posts", {
    "query": "competitor_name vs OR competitor_name alternative OR switching from competitor_name",
    "count": 100
})
```

**Sentiment scoring:**
```
For each