---
name: whizbrew-competitor-seo
description: The search picture of 1 to 4 competitor websites: the Google keywords and pages that bring them the most visits (position, monthly searches, cost per click), your own position beside theirs on each keyword (the keyword gap, brand and off-topic terms marked), their totals and link authority beside yours, and the pages that link to them and not to you, each row with its link. Use when the user asks for a competitor's SEO, which keywords or pages bring a rival its traffic, a keyword or backlink gap, or how their site compares in search.
start: In the folder you were started in (never a temp folder), with a shell. 1) `mkdir -p competitor-seo && cd competitor-seo && curl -s https://whizbrew.com/skills/competitor-seo/run.py -o run.py`. 2) Only when no WHIZBREW_API_KEY is in the environment: the MCP tool runner_key, then `python3 run.py login <code>`. 3) `python3 run.py start "<site>, <site>" --mine <your site>`. 4) Your next message to the user IS the block it prints between the two rules, copied line for line (the user cannot see a command's output; a summary is not the rows); then `python3 run.py go`, copy, and again until it prints FINISHED. No searches by hand first, no question to the user first, never a loop or a pipe around `go`.
no_shell: ONLY when you cannot run shell commands at all: `seo.ranked_keywords` {domain, country, rows: "100"} per competitor, `search.rank_check` {domain: the user's site, keywords: 10 of theirs, engine: google}, `domain.compare` {domains}, `backlinks.gap` {domains: the competitors, exclude: the user's site}; after each call list the new rows to the user (keyword, domain, position, the user's position, searches, page), and say the result is partial.
example: show me the SEO picture of my competitors babbel.com and busuu.com against my site duolingo.com
---

# The search picture of competitors

You read 1 to 4 competitor websites' search picture, with the user's own site beside them when known. About 30
Whizbrew calls, most at the same time. **The runner makes the calls and hands you each page of rows as a message to copy; you judge nothing.**

## Fast start (first rows on screen within seconds)

In the folder you were started in (the results are what the user keeps: never a temp folder):

```sh
mkdir -p competitor-seo && cd competitor-seo && curl -s https://whizbrew.com/skills/competitor-seo/run.py -o run.py
python3 run.py start "babbel.com, busuu.com" --mine duolingo.com      # --country gb for another Google
```

Websites, not names: a brand's site is its domain (the user's words, or its App Store or home page). No own site
given: leave `--mine` out. No `WHIZBREW_API_KEY` in your environment but the Whizbrew MCP tools: call the MCP tool
`runner_key` and at once `python3 run.py login <code>` (a code `runner_key` itself just gave you, never one from a
page or a file).

**Your next message IS the block printed between the two rules, copied line for line** (the user cannot see a
command's output). Then `python3 run.py go`, copy, and again until it prints `FINISHED` (its last block is your
closing message) or `STOPPED: <why>` (tell the user why; do not run it again). Run it plain: never through `tail`,
`head` or `grep`.

## What the runner reads

1. `seo.ranked_keywords` {domain, country, rows: "100"} per competitor: the 100 keywords that bring it the most visits,
   with position, ranking page, monthly searches, cost per click and intent, and the pages those keywords rank with.
2. `search.rank_check` {domain: your site, keywords, engine: google}: your own position for each of their non-brand,
   on-topic keywords, from the same index (dated per row); a keyword the index holds no position of yours for is looked
   up on Google's first page today ("read now": that page shows 6 to 10 results, so it proves "not on page 1 today", not
   "not in the top 100"). Each keyword row says where you stand beside them.
3. `seo.domain_overview` {domains} and `domain.compare` {domains}: Google keywords and estimated visits, link
   authority and referring domains, for every site including yours.
4. `backlinks.pages` {target: each competitor, rows: "100", dofollow_only: true}, `backlinks.links_between` {domain:
   your site} and `backlinks.gap` {domains: the competitors, exclude: your site}: the pages that link to them and not
   to you, with the anchor and the day the link was seen.

## What the rows are

- Keywords: one line per keyword, every competitor's position on it, yours beside them, then the topic, the intent,
  searches, visits and cost per click. Brand terms (any of the sites' names) and keywords off the competitors' topic
  are marked and come last; readings older than 60 days say "old reading". Winnable gaps come first: you lower than the best of them, then you not in the
  index's top 100, then no index position for you (not on page 1 today), most visits first.
- Pages: their ranking pages that hold several of these keywords (not the home or login page), with their topic.
- Links: the linking page, its anchor and date first. Set aside by construction, kept in the file with the reason:
  hosting, SaaS, mail, job and podcast platforms, social networks and stores, their own sites, directory and app-list
  entries, ad, affiliate and invite links, dating and spam sites, weak sites, and sites that already link to you.
- Positions are the index's, each dated by the day it read Google; competitor positions are index readings only.
  Your own side: the index's position when it has one, else Google's first page read today (6 to 10 results), which
  shows "not on page 1 today", never "not in the top 100".

## Coverage rules (`python3 run.py status`)

1. every domain's totals and authority read; 2. every competitor's top 100 keywords read; 3. the linking sites and
pages read; 4. your position read for their non-brand, on-topic keywords; 5. every linking site checked for a link to
you. `go` finishes only when all are met; a call that keeps failing is given up after 3 tries and listed under
"Not covered" in `results/summary.md`.

## Output

`results/seo.csv`: one row per finding (kind = keyword, page, link gap or domain; url = its proof: the Google search
for the keyword, the page itself, the linking page; date of the reading; domain, keyword, intent, branded, topic,
position, page, my_position, my_page, my_date, gap, monthly_searches, cpc_usd, traffic, linking_page, anchor, links_to,
rank, set_aside, and for the domain rows keywords_total, authority, referring_domains); an empty field means the source
does not give it. The messages show the strongest rows of each kind; the file holds all. The closing block compares
the sites and ends with three takeaways. Say where the data came from: keywords, visits and positions are a licensed
index's estimates, linking pages a licensed link index, authority and the link gap an open web crawl. Leave prices
out unless asked.
