competitive-pricing-intel

name	competitive-pricing-intel
description	Monitor competitor pricing pages over time — capture current plans/tiers/feature-gating, pull Wayback history, and diff against the last snapshot to detect price changes, tier restructuring, model shifts, and gating moves. Produces a living pricing comparison matrix plus change alerts. Keyless backbone; the agent normalizes and writes the report.
metadata	{"version":"1.0.1","category":"competitive-intel","type":"composite"}

Competitive Pricing Intel

Composite: deterministic scripts capture live + historical pricing pages and diff snapshots across runs; you, the agent, normalize heterogeneous pricing, compute ICP-scenario cost, and write the report. Two modes: first_run (baseline) and recurring (diff).

When to use

• "What are our competitors charging?" / "Has [competitor] changed pricing recently?"
• "Build a pricing comparison matrix." / "Monitor competitor pricing for changes."
• Runs standalone or as a recurring sub-routine inside competitor-monitoring-system.

How to run

1. Live capture (per competitor)

python3 ${SKILL_DIR}/scripts/fetch_pages.py \
  --url https://competitor.com/pricing --output ${WORKSPACE}/acme_pricing.json

Interactive pricing (monthly↔annual toggles, currency/region, calculators) is JS — render it:

npx playwright install chromium   # first run only
node ${SKILL_DIR}/scripts/render_page.mjs \
  --url https://competitor.com/pricing --wait 5000 \
  --output ${WORKSPACE}/acme_pricing_rendered.json

2. Historical check (Wayback)

python3 ${SKILL_DIR}/scripts/wayback_fetch.py \
  --url https://competitor.com/pricing --snapshots 3 \
  --output ${WORKSPACE}/acme_pricing_history.json

Each snapshot includes prices_found and a text excerpt so you can spot tier/price moves.

3. Announcement research

Use your own web search for "[competitor]" "new pricing" OR "updated plans" and site:reddit.com [competitor] "price increase". Apify Reddit actor only if you need depth (see Notes).

4. Normalize + analyze (you, the agent — no script)

From the live + historical JSON, build the normalized matrix: plan names, monthly/annual prices, limits, add-ons, free tier, enterprise trigger, model. Reduce per-seat vs usage vs flat to an ICP-scenario effective cost before comparing. Classify each competitor's packaging strategy.

5. Diff across runs (recurring mode)

Write the fields you want to track as a flat JSON object (e.g. {"starter_price":"$29/seat/mo","tiers":3,"free_tier":true}), then:

# recurring: compare to last saved snapshot, then save the new one
python3 ${SKILL_DIR}/scripts/snapshot_store.py diff --entity acme \
  --input ${WORKSPACE}/acme_fields.json \
  --store ${WORKSPACE}/supabase/pricing_history.csv
python3 ${SKILL_DIR}/scripts/snapshot_store.py save --entity acme \
  --input ${WORKSPACE}/acme_fields.json \
  --store ${WORKSPACE}/supabase/pricing_history.csv

diff returns added / removed / changed (with prev→current). Rate each change by severity (price, tier, gating, model, free-tier). first_run: true means baseline only — do not fake a diff.

6. Report (you, the agent)

Write pricing-comparison-[YYYY-MM-DD].md: change-alert section, normalized matrix, ICP-scenario cost table, feature-gating comparison, packaging summary, recommendations.

Outputs

• ${WORKSPACE}/*_pricing.json, *_pricing_history.json — captured signal.
• ${WORKSPACE}/supabase/pricing_history.csv — durable snapshot history for diffing.
• ${WORKSPACE}/pricing-comparison-[date].md — the report (your synthesis).

Credentials / env

• Required: none — fetch / render / Wayback / snapshot-store are all keyless.
• Optional: SUPABASE_URL/SUPABASE_KEY or AIRTABLE_API_KEY for durable history in a shared store (the CSV store works offline without them); APIFY_API_TOKEN for deeper Reddit reaction mining; DATAFORSEO_LOGIN/DATAFORSEO_PASSWORD or SERPER_API_KEY for the pricing-announcement / reaction searches (if set -> paid SERP; if not -> the agent's own web search, the default).

Notes & edge cases

• Pricing pages change infrequently — schedule monthly.
• Wayback may lack recent snapshots; if so the first run is the de-facto baseline and change detection starts next cycle. The script returns a "first run — no prior snapshot" note rather than faking a diff.
• Normalize before comparing: per-seat vs usage vs flat must be reduced to an ICP-scenario effective cost or the matrix misleads.
• Some vendors localize prices by IP — capture the buyer-relevant region/currency with the renderer.
• Apify degrade (when set): curl -s "https://api.apify.com/v2/acts/<actor>/run-sync-get-dataset-items?token=$APIFY_API_TOKEN" -d '{...}'.