Riot starts banning Valorant accounts for tools that automate the agent pick

A streamer's queue is public by construction: viewers can see when the search starts and roughly when the pick screen will appear. That surface sits outside the game client, and it is half of what Riot has now declared bannable ^[1]. The other half, a tool that locks an agent portrait faster than a person can, sits inside the client but inside a menu rather than a match. Bundling both into one sentence is tidy for patch notes and awkward for enforcement, because the evidence lives in different places: one is input automation you can look for on a machine, the other is a correlation between a match and a broadcast Riot does not operate.

The published policy runs 35 words across two sentences ^[8]. It does not say what counts as too fast, what a first offence costs in days, or how a player contests it ^[2]. The same notes are more specific about smoke dissipation timing for five named agents ^[11] and about narrowing sound spray audibility so nearby teammates can still hear each other ^[10]. That ordering is normal in a housekeeping patch. It is also how a policy ends up being defined in practice by whoever answers the tickets. Per Dexerto, this is the second enforcement expansion outside the shooting itself this year, after a January push on multi-factor authentication aimed at smurfing and cheating ^[6].

Now the incentive. From September 22, each piece of Agent Gear is tied directly to how much you play the associated agent ^[5]. Progression that pays per agent makes the pick phase worth more to exactly the player who was already tempted to automate it ^[3]. Riot is shipping both in consecutive patches, which is a coordination problem rather than a contradiction, and a team copying the ban without noticing the progression half inherits the enforcement load without understanding where the demand came from.

For anyone asked to do this in their own game, the useful split is whether you can observe the mechanism or only the outcome, crossed against whether your penalty is reversible. An observable mechanism paired with escalating bans is defensible, as long as you publish the threshold. When you can only see the outcome, reversible friction, a randomized pick timer, or a queue that hides who you are about to play, fits most teams better, because it removes the advantage instead of adjudicating intent. Pairing outcome-only detection with permanent bans creates a support headcount problem that never really goes away. Riot's wording says tools, which is a mechanism claim ^[1]; whether the detector can hold that line is the part no one outside Riot can check.

If you cannot name the artifact on the machine, the pick phase itself is the thing to change, not the ban. The tradeoff is that friction taxes everyone, including the one-trick main who is not cheating and now waits an extra beat, while a ban only taxes the players your detector selected. Which cost you would rather carry depends on the false-positive rate you can state out loud. Most teams cannot state one.