Check guide ยท Platforms
Which platforms care about which tells
Blowhorn works on seven platforms. Their defences are not the same shape: one reads your browser down to its fonts, another barely runs JavaScript at all. This page ranks them by bot-detection sensitivity and maps each check in the guide to the platforms where it plausibly matters, with the evidence labelled for what it is.
How to read the evidence
Claims are tagged. [Vendor]: the platform states it in its rules, help pages or API terms. [Reported]: researchers or practitioners documented it, in some cases disputed by the platform. [Inference]: our reasoning from how the platform is built, labelled because it is not documented fact. A Low rating means no public evidence a tell matters there. It never means proof that nobody looks.
The ranking
| Rank | Platform |
|---|---|
| 1 | |
| 2 | X |
| 3 | |
| 4 | GitHub |
| 5 | Slack |
| 6 | Bluesky |
| 7 | Hacker News |
Most sensitive first. The pattern underneath the order: the higher a platform ranks, the more its defences move from the account (what you do) toward the browser and the connection (what you are). The bottom of the table is not defenceless; it is defended at a different layer entirely.
1. LinkedIn
The most sensitive platform in this set, and the only one whose defences reach deep into the browser. LinkedIn's rules ban automation software outright and name account restriction as the consequence [Vendor]. Independent researchers reported in 2026 that its pages probe for thousands of installed Chrome extensions and attach an encrypted device fingerprint, built from dozens of hardware and software characteristics, to session requests [Reported; LinkedIn disputes parts of that characterisation]. Treat every browser-side tell as live here.
2. X
X permits automation through its API and forbids scripting its website, with permanent suspension as the stated penalty [Vendor]. In practice its defences are a spam model: IP reputation, velocity, and behaviour, with Arkose challenges at the moments that matter (unlocks, sensitive actions) [Reported]. The network and behaviour families decide most outcomes; deep browser fingerprinting is less documented here than on LinkedIn.
3. Reddit
Reddit fights at the network edge. Unregistered automation meets a flat "blocked by network security" wall, and the reported triggers are request-shape problems: datacenter and VPN addresses, missing browser headers, inhuman timing [Reported]. Accounts with history fare better than fresh ones, but the first question Reddit asks is about your connection, not your browser's canvas.
4. GitHub
GitHub expects automation and gives it an API, published rate limits (including content-creation caps that also bind the web UI), and an acceptable-use ban on bulk inauthentic activity [Vendor]. Browser tells are nearly irrelevant here; what gets accounts flagged is what they do in bulk, not what their browser looks like. The tells that matter are the behavioural ones, read at account scale.
5. Slack
Slack's trust model is the workspace, the app and the token. It publishes tiered rate limits and says it detects apps acting spammy and quiets them [Vendor]. There is no public browser-fingerprinting surface to speak of; a session is judged by its credentials and its conduct inside a workspace that admitted it.
6. Bluesky
An open protocol that expects bots and labels many of them good. Spam is defined behaviourally in its guidelines, rate limits are published per IP and per account, and enforcement runs through labelling and moderation rather than device signals [Vendor]. Browser tells barely register; network citizenship and behaviour carry what little weight exists.
7. Hacker News
The least sensitive to browser tells, for a structural reason: the site is server-rendered with almost no JavaScript, so most browser tells have nothing to attach to. Enforcement is real but orthogonal: IP throttling, voting-ring detection and shadowbans, confirmed in on-record moderator statements [Reported]. The network family (which IP, how it behaves) is the whole game.
The matrix
Every check against every platform. Follow a check name for its full page, including the per-platform reasoning.
| Check | X | GitHub | Slack | Bluesky | Hacker News | ||
|---|---|---|---|---|---|---|---|
| Request profile | H | H | H | M | L | L | L |
| Proxy indicators | M | H | H | M | L | L | M |
| Datacenter pattern | M | H | H | L | L | L | M |
| VPN / proxy / Tor intelligence | M | M | H | M | L | L | M |
| Managed bot score | M | M | M | L | L | L | L |
| TLS fingerprint | M | M | M | L | L | L | L |
| Browser data | H | M | M | L | L | L | L |
| Browser fingerprint | H | M | M | L | L | L | L |
| Fingerprint surfaces | H | M | L | L | L | L | L |
| Performance memory pings | M | L | L | L | L | L | L |
| Worker consistency | L | L | L | L | L | L | L |
| WebRTC network clues | L | L | L | L | L | L | L |
| Latency | L | M | M | L | L | L | L |
| Remote-control tells | H | H | M | M | L | L | L |
| Behavioural classification | H | H | M | M | M | M | M |
| Optional human check | H | H | M | M | L | L | L |
Two rows are Low across the board: Worker consistency and WebRTC network clues. That is an honest gap in the public record for these seven platforms, not an oversight in this table; both tells remain standard inputs to commercial bot-management products, so they are the rows most likely to change silently. Researched October 2026; treat the whole matrix as a dated snapshot.