/pricing-finder

name: pricing-finder
description: 'Tell it what your product is (URL or description) and it finds 5 competitors globally, fetches their actual pricing pages, extracts every tier and price point, and returns a complete pricing intelligence report: the dominant pricing model in your space, a benchmark price table, feature gate analysis, competitive positioning map, and a concrete recommended pricing strategy for your product. Use when asked to research competitor pricing, find pricing benchmarks, decide how to price a product, understand pricing models in a space, or build a pricing strategy.'
compatibility: [claude-code, gemini-cli, github-copilot]

Pricing Finder

Tell it your product URL or description. It finds 5 competitors, fetches their actual pricing pages, and returns a complete pricing intelligence report: dominant model in your space, benchmark price table, feature gate analysis, positioning map, and a concrete pricing recommendation for your product.

**Zero required API keys.** Runs entirely on free pip dependencies. Optional API keys improve quality.

---

**Zero-hallucination policy:** Every price point, tier name, and feature gate in the output must trace to fetched pricing page content or a DuckDuckGo search snippet. This applies to:

• Competitor prices: extracted verbatim from fetched page content only
• "Contact Sales": recorded as-is, never estimated or replaced with a number
• Tier names: copied exactly from the page, not paraphrased
• Feature lists: extracted from page content, not inferred from product knowledge
• Positioning observations: derived from the benchmark table data only

---

Common Mistakes

| The agent will want to... | Why that's wrong | |---|---| | Fill in "Contact Sales" with an estimated price | Never estimate enterprise pricing. Record it as "Contact Sales" exactly. | | Use training knowledge for competitor prices | Every price must trace to fetched page content or a search snippet. | | Skip the competitor confirmation step | Always show discovered competitors and wait for confirmation. Wrong competitors = wrong benchmarks. | | Recommend a price without referencing benchmark data | Every price recommendation must cite a specific number from the benchmark table. | | Mark a page as high quality when content < 500 chars | < 500 chars means the page was not fetched -- mark data_quality as 'low' and use search snippet fallback. | | Use em dashes in output | Replace all em dashes with hyphens. |

---

Read Reference Files Before Each Run

cat references/pricing-models.md
cat references/extraction-guide.md
cat references/positioning-guide.md

---

Step 1: Setup Check

echo "TAVILY_API_KEY:    ${TAVILY_API_KEY:+set (search quality enhanced)}${TAVILY_API_KEY:-not set, DuckDuckGo will be used (free)}"
echo "FIRECRAWL_API_KEY: ${FIRECRAWL_API_KEY:+set (JS rendering enhanced)}${FIRECRAWL_API_KEY:-not set, requests+BS4 will be used (free)}"
echo ""
python3 -c "from ddgs import DDGS; import requests, bs4, html2text; print('Dependencies OK')" 2>/dev/null \
  || echo "ERROR: Missing dependencies. Run: pip install ddgs requests beautifulsoup4 html2text"

**If dependencies are missing:** Stop immediately. Tell the user: "Missing Python dependencies. Run this to install them: `pip install ddgs requests beautifulsoup4 html2text` -- all free, no accounts needed. Then try again."

**If only API keys are missing:** Continue. DuckDuckGo and requests+BS4 are the free defaults.

Derive product slug:

PRODUCT_SLUG=$(python3 -c "
from urllib.parse import urlparse
import sys, re
url = 'URL_HERE'
if url.startswith('http'):
    host = urlparse(url).netloc.replace('www.', '')
    print(host.split('.')[0])
else:
    print(re.sub(r'[^a-z0-9]', '-', url[:30].lower()).strip('-'))
")
echo "Product slug: $PRODUCT_SLUG"

---

Step 2: Parse Input

Collect from the conversation:

• `product_url`: the URL to fetch (required, unless user pastes a description directly)
• `geography`: optional -- US / Europe / India / global. Default: US

**If the user provides only a pasted description (no URL):** Skip Steps 3 and 4. Go directly to Step 4 (product analysis) using the pasted text as `product_content`. Set `page_source` to `user_description` and note in `data_quality_flags`.

**If neither URL nor description:** Ask: "What is the URL of your product or startup? Or paste a short description: what it does, who it's for, and what makes it different."

---

Step 3: Fetch Product Page

**Primary: Firecrawl (if FIRECRAWL_API_KEY is set)**

curl -s -X POST https://api.firecrawl.dev/v1/scrape \
  -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "URL_HERE", "formats": ["markdown"], "onlyMainContent": true}' \
  | python3 -c "
import sys, json
d = json.load(sys.stdin)
content = d.get('data', {}).get('markdown', '') or d.get('markdown', '')
print(f'Fetched via Firecrawl: {len(content)} characters')
open('/tmp/pf-product-raw.md', 'w').write(content)
"

**Fallback: requests + BS4 (free, always available)**

python3 << 'PYEOF'
import requests, html2text, random

USER_AGENTS = [
    "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
]
headers = {"User-Agent": random.choice(USER_AGENTS), "Accept": "text/html,application/xhtml+xml;q=0.9,*/*;q=0.8"}
resp = requests.get("URL_HERE", headers=headers, timeout=20, allow_redirects=True)
converter = html2text.HTML2Text()
converter.ignore_images = True
converter.body_width = 0
content = converter.handle(resp.text)[:8000]
print(f'Fetched via requests+BS4: {len(content)} characters')
open('/tmp/pf-product-raw.md', 'w').write(content)
PYEOF

**Checkpoint:**

python3 -c "
content = open('/tmp/pf-product-raw.md').read()
if len(content) < 200: