Blog
How to moderate a CS2 skin trading Discord (where every scammer wants in)
Moderating a CS2 skin trading Discord is less like running a chat room and more like policing a marketplace: the skins changing hands carry real-world value, so the server attracts professional scammers the way money always does. The defenses that work are structural: verified middleman systems and reputation channels with strict rules, hard link hygiene (phishing sites are the number one attack), instant-ban lines for impersonation, and a watching layer that can spot scam patterns in polite-looking messages, because the defining feature of trading scams is that they never use words a filter would catch.
Why trading servers are scammer magnets
CS2 skins are one of gaming's oldest real-money economies: rare items trade for the price of a used car, and the wider skin market is commonly valued in the billions. Wherever that value flows, scammers follow, and Discord trading servers are their richest hunting ground because the platform layer (Steam's own trade protections) does not reach into your chat. The scam happens in the conversation: the trust-building, the fake link, the impersonation. Which means the server owner is, functionally, the marketplace's regulator, and your members' losses land on your reputation whether you caused them or not.
Know the classic scams (they rotate, the shapes repeat)
The recurring shapes every trading mod should recognize: phishing links dressed as trading sites, gambling sites, or screenshot tools (the number one vector; the site harvests the victim's Steam login). Impersonation: fake middlemen, fake staff accounts one character off from a real name, fake bot accounts. API-key and trade-offer manipulation, where a compromised account's real trades get swapped with lookalike bot offers. Fake rep: scammers farming vouches in small servers, or presenting screenshotted "reputation" that a real system would expose. And overpayment or item-switch moves inside the trade window itself.
The common thread: none of these messages read as hostile. They read as friendly, helpful, and professional, which is exactly why keyword filters are useless against them.
Structure the marketplace: middlemen, rep, and trade channels
The strongest anti-scam tools are structural. Run a verified middleman system: a small, vetted staff-adjacent group with an unmistakable role, and a standing rule that legitimate middlemen never DM first. Keep a reputation system with integrity rules: vouches only in the rep channel, only for completed trades, with fake-vouch farming an instant ban. Confine trading to designated channels with a pinned format (item, price, payment method), which makes anomalies visible. And publish a scam-response protocol: report channel, evidence expectations, and a public tag or ban list so repeat offenders cannot simply rejoin quietly.
The rules that need zero tolerance
Trading communities run on trust, so the rules protecting trust get the harshest ladder: impersonating staff or middlemen, posting phishing links, and fake-vouch schemes should be instant permanent bans, no first-offense warning. Link hygiene deserves its own hard rule: only allowlisted trading sites, no link shorteners, no "check my inventory on this site" links. New-account friction helps too: account-age gates and a probation role before posting in trade channels, because scam accounts are usually young and disposable.
The watching layer: scams are a context problem
Here is the moderation problem underneath all of it: scam behavior is a pattern, not a keyword. A friendly newcomer offering to help, a link that looks almost right, a "middleman" whose name is one letter off, urgency pressure in an otherwise polite DM invitation: each message alone looks fine, and the pattern only reads as a scam in context. That is precisely the tier our AI moderator Staffd works in (disclosure: we make it): it learns your server's rules and structures (your real middleman roles, your link policy, your rep system), reads messages in context, remembers member history, and flags the pattern to your staff channel with reasoning, where a human makes the call. Filters guard against words; a trading server needs something that guards against intent.
The full stack for a trading community: structural rules and verified middlemen, hard link hygiene, account-age gates, Discord AutoMod for the crude layer, and a context-reading watcher on top. Your members are trusting your server with real money. The servers that take that seriously are the ones that grow.