---
name: whizbrew-brand-monitor
description: What people say about a brand right now, on TikTok, Instagram, X, YouTube, Reddit, Hacker News, Bluesky, Threads and the news: each post with its link, time, author and reach, its type (complaint, question, praise, news, promo) and whether it needs a reply, most urgent first, the brand's own posts and namesakes set apart, and the comments under its most seen videos that ask or complain. Use when the user asks what people say or think about a brand, product or app this week, for brand mentions, social listening, buzz, sentiment or share of voice, or to monitor a competitor's brand.
start: In the folder you were started in (never a temp folder), with a shell. 1) `mkdir -p brand-monitor && cd brand-monitor && curl -s https://whizbrew.com/skills/brand-monitor/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 "<brand>" --site <its website> --about "<what it is, in a few words>"`, with `--days <N>` when the user named a period, or `--since <UTC time, e.g. 2026-11-20T09:00>` when the user names a start time. 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: `social.brand_mentions` {query: the brand, limit: 10}, `tiktok.search` {query: the brand, sort: newest, time_range: week}, `x.search` {query: the brand, sort: newest, time_range: week}, `news.search` {query: the brand, days: 7}; keep only posts that are about the brand (not a namesake), list each to the user (platform, author, date, text, link) after each call, and say the result is partial.
example: what are people saying about Notion this week on social media?
---

# What people say about a brand

You read what people post about one brand in a recent window, across the platforms a visitor can search. About 25 to
40 Whizbrew calls. **The runner makes the calls and hands you each page of mentions as a message to copy; you judge
nothing.**

## Fast start (first mentions on screen within about half a minute)

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

```sh
mkdir -p brand-monitor && cd brand-monitor && curl -s https://whizbrew.com/skills/brand-monitor/run.py -o run.py
python3 run.py start "Notion" --site notion.so --about "note-taking, docs and wiki app for work and study"
```

- The brand as people write it; `--site` its website (its own accounts are read from it and set apart);
  `--about` what it is in a few words (it tells the brand apart from everything else of the same name).
- `--days` the window (default 7); `--since 2026-11-20T09:00` only posts from that UTC time on. `--words "Notion AI"`: up to 2 more names people use.
- `--distinct` only when the name means nothing else (no judge needed). `--subreddits "a, b"`: where its users talk.
- `--handles "brandhq, brand_app"`: its own accounts when it has no website. `--platforms tiktok,x,web`: only those
  (tiktok, instagram, x, youtube, reddit, hackernews, web = the open web and news; default all).
- 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. Its own accounts: `web.social_links` {url: the site}.
2. For every name, in the window: `tiktok.search` (newest, and TikTok's order), `ig.search_posts` (newest), `x.search`
   (newest, and X's Top), `youtube.search` (newest videos), `hackernews.search`, `social.brand_mentions` (Bluesky,
   Threads, YouTube, Hacker News, Lemmy, Mastodon, news) and `news.search`; Reddit: `reddit.subreddit_search` for the
   brand's own subreddits, then `reddit.search` in them.
3. Is it about the brand: a post by its own account, or naming its handle or site, is. A post that only names it bare
   (or its hashtag) goes to the post judge `creators.check_posts` {app, about, posts}: kept when the judge is at least
   97% sure, or sure and the post itself says what the brand is (a word of `--about`). A post in the brand's own
   subreddit is about it. Posts the judge read and could not place are left out and counted; posts it never answered for
   are kept, last, marked "to check". A post that only names the brand in a list of other tools is left out and counted.
4. The comments: `tiktok.comments` under its three most seen TikToks, `youtube.comments` under its two most seen
   YouTube videos (only posts surely about it); kept only when they ask, complain, or say something about the brand.
5. What each post is, from its own words: type (complaint, question, praise, news, promo, post), tone (negative, positive
   or empty = not judged), needs_reply, tagged_us (it tags the brand's account), why_type (the words that decided).

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

1. every name searched on every platform asked, in the window; 2. every post that names it bare judged; 3. the
comments under its most seen videos read. `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/mentions.csv`: one row per post or comment (url = the post itself, the proof; date = UTC time "YYYY-MM-DD HH:MM",
or the day when the source gives no hour; type, tone, needs_reply, tagged_us, platform, author, author_followers, views,
likes, comments, text, about_it = why it counts, why_type, sponsored, own_account, found_by). The messages show the most
urgent first: a leak, breach or outage story, then complaints, questions, praise that tags the brand, news, the rest by
reach, promo and posts to check last. The closing block starts with the story to watch and "reply to these first", then
counts by type and tone, per platform, a day, and what was left out. The runner writes no suggested replies. It is a sample of what each platform's own search shows, not a
count of every post; Facebook and LinkedIn give no public post search. Leave prices out unless asked.
