Competitor Intel

Competitor Intel

Composite and reusable primitive: deterministic scripts collect public signal per competitor across three layers (data collection, content tracking, optional monitoring); you, the agent, write each profile, synthesize the landscape, and recommend actions.

When to use

• "Research [competitor]." / "Build a competitor profile for [company]."
• "What are our competitors doing?" / "Competitive landscape analysis."
• Baseline engine invoked per-competitor by competitor-monitoring-system; shares its review/social/pricing legs with battlecard-generator and company-current-gtm-analysis.

How to run

Loop per competitor (2–5). Pick quick (website + serp + one review pass) or deep (adds social + full review/ad mining) depth.

1. Company / product / positioning pages

python3 ${SKILL_DIR}/scripts/fetch_pages.py \
  --url https://competitor.com https://competitor.com/about \
        https://competitor.com/pricing https://competitor.com/product \
        https://competitor.com/integrations https://competitor.com/customers \
  --output ${WORKSPACE}/acme_pages.json

Use your own web search for [company] funding, [company] founders, [company] launch 2026, and [company] vs / alternatives pages. Re-fetch JS-rendered pages with node ${SKILL_DIR}/scripts/render_page.mjs --url <u> --output ....

2. Content & marketing

python3 ${SKILL_DIR}/scripts/fetch_feed.py --url https://competitor.com/blog \
  --output ${WORKSPACE}/acme_blog.json     # cadence, authors, recent topics

Find social handles via web search. For founder/company LinkedIn activity use render_page.mjs (one-off) or PHANTOMBUSTER_API_KEY (scale).

3. Customer evidence (reviews)

Render the G2/Capterra review page (anti-bot, JS):

npx playwright install chromium   # first run only
node ${SKILL_DIR}/scripts/render_page.mjs \
  --url "https://www.g2.com/products/<acme>/reviews" \
  --selector "[itemprop='review']" --output ${WORKSPACE}/acme_g2.json

Pull rating + praise/complaint themes. Apify fallback on hard block (see Notes).

4. Optional social depth (deep only)

Reddit/X/LinkedIn at depth → APIFY_API_TOKEN actors or PHANTOMBUSTER_API_KEY (LinkedIn). Without keys, your own web search covers the basics.

5. Profile + synthesize (you, the agent — no script)

Per competitor write: overview, positioning, product, content/marketing, customer evidence, signals, strengths/weaknesses-vs-you, your opportunity. Then synthesize a landscape summary (positioning map, content comparison, feature comparison, key takeaways) and recommended actions — ground every recommendation in a collected signal. Dedup across sources; an item on both Reddit and X is higher signal — flag it.

6. Optional ongoing monitoring

Write the fields to track as flat JSON, then diff + save across runs:

python3 ${SKILL_DIR}/scripts/snapshot_store.py diff --entity acme \
  --input ${WORKSPACE}/acme_track.json --store ${WORKSPACE}/supabase/competitor_history.csv
python3 ${SKILL_DIR}/scripts/snapshot_store.py save --entity acme \
  --input ${WORKSPACE}/acme_track.json --store ${WORKSPACE}/supabase/competitor_history.csv

Outputs

• ${WORKSPACE}/*_pages.json, *_blog.json, *_g2.json — collected signal per competitor.
• ${WORKSPACE}/competitor-profiles-[date].md — profiles + landscape + actions (your synthesis).
• ${WORKSPACE}/supabase/competitor_history.csv — monitoring history (if step 6 used).

Credentials / env

• Required: none — baseline research (fetch / feed / render) is free; synthesis is the agent.
• Optional: APIFY_API_TOKEN (Reddit/X/LinkedIn/review depth); PHANTOMBUSTER_API_KEY
  • LinkedIn cookie (LinkedIn at scale); SUPABASE_URL/SUPABASE_KEY or AIRTABLE_API_KEY (durable profile history for monitoring); DATAFORSEO_LOGIN/DATAFORSEO_PASSWORD or SERPER_API_KEY for the funding/founders/alternatives searches (if set -> paid SERP; if not -> the agent's own web search, the default).

Notes & edge cases

• Default to the free backbone; reach for Apify/Phantombuster only for social depth or hostile review sites.
• quick = website + serp + a single review pass; deep adds social + full review/ad mining.
• Ground every recommendation in a collected signal — no generic advice.
• Apify degrade (when set): curl -s "https://api.apify.com/v2/acts/<actor>/run-sync-get-dataset-items?token=$APIFY_API_TOKEN" -d '{...}'.