Features
What youactually get
A report that shows its work, a badge on X, a public URL, and the same JSON over the API. Not three icon cards.
Every scan lands here: a score, a tier, and a story.
01
The three reasons, with the evidence
The score is not a black box. Each report names the checks that fired, shows the numbers, and puts the posts next to the claim.
How scoring works
02
Everything else we checked
Five groups under the reasons: when it posts, what it posts, how it talks, who it follows, and the profile itself. Expand a group for the checks that passed, the ones we couldn't run, and the charts.
See the groups on the sampleWhen it posts
3 fired · 5 passed
- Posted in 22 of 24 hours last week — longest break 3 hours.
- Quiet for days, then a burst of 40 posts in one afternoon.
What it posts
2 fired · 4 passedHow it talks to people
2 fired · 3 passedWho it follows and who follows it
1 fired · 3 passedThe profile itself
1 fired · 2 passed03
One scan
Every plan collects the same sample: posts, replies under the top posts, and follower, following, and amplifier samples, with AI checks. Same rules. Same score meaning.
Plans and limits
04
See it on X, before you reply
The extension puts a small badge next to accounts already scored in your feed. One badge per account. Grey until there is a reason. Scanning shows up as a wait, not a fake score.
The extensionThis is the way. Everyone needs to see this. #crypto #alpha
Our peer-reviewed paper on Arctic ice loss is out.
The DNC just proved it again. Retweet if you're done with it.
Thread: what the council vote actually changes (1/6)
05
The three reasons, in the tab
Tap the badge for the score, the tier line, and the three reasons with their numbers — the same words the full report uses. Then jump to the report, hide the posts, or never flag that account.
How the popover readsVery likely a bot or a run account
A megaphone, not a person.
- Posted in 22 of 24 hours last week — longest break 3 hours.
- The same post, word for word, 4 times. Only 1 in 20 posts is original.
- 80 replies out, 2 back — 6 in 10 don't fit the post they're under.
Sample: 120 posts and 80 replies collected, 81% of target
BotShark's read of public posting patterns — not a statement of fact about the person.
06
On a profile, under the bio
Open an account and the verdict sits where people already look. Score, tier, headline, and a link to the full report — not a toast that vanishes.
Profile and threadBreaking narratives | alpha drops | DM for promo rates
Very likely a bot or a run account
A megaphone, not a person.
Full reportScored 3 days ago · not a statement of fact
07
A public URL for every report
Finished reports live at botshark.io/u/handle. Send the link. Anyone can read the reasons on a moderate or high report; one with a low score, or too little data to score, opens only for signed-in visitors. Anyone can dispute it if we got it wrong.
Open the sample URLCoordination Watch@coordination_watch
Very likely a bot or a run account
A megaphone, not a person.
- Posted in 22 of 24 hours last week
- The same post, word for word, 4 times
- 80 replies out, 2 back
08
The same report over the API
Personal and Pro mint a key under Settings. The JSON is the public report shape — score, tier, confidence, and the checks that fired. Not a different, vaguer product.
Plans and limitscurl -s -H "Authorization: Bearer $BOTSHARK_API_KEY" \
"https://app.botshark.io/v1/handles/coordination_watch/latest" \
| jq '{score, tier, confidence}'
{
"score": 87,
"tier": "very_likely",
"confidence": 0.81
}09
Follower audits, for the ones around it
Personal and Pro can sample who follows a handle and score the cohort. You see which of those accounts look automated — not just the subject.
Personal and Pro| Handle | Score | Read |
|---|---|---|
| @alpha_drop_bot | 91 | likely automated |
| @promo_relay | 84 | likely automated |
| @patriot_signal | 72 | some signs |
| @arcticdatalab | 12 | looks like a person |
| @maria_k_writes | 18 | looks like a person |
38% of the sampled followers look automated. Personal and Pro.