Where the term appears in this channel Slash commands show up in the channel's material as the typed-command pattern used while interacting with AI coding tools and agents — short inputs prefixed with a slash that trigger a built-in action rather than a plain message to the model. During the channel's live model tests, agent workflows are a recurring trial: the hosts hand models real work — pull-request triage, repo maintenance, and document tasks — and part of judging a model is how it handles the command-driven interface around it, not just raw prompt completion. How to think about it A slash command is best understood as a shortcut for intent: instead of writing a long natural-language instruction, the user types a compact command and the tool maps it to a defined behavior. In agent-heavy work, these commands sit at the boundary between the human and the model — they shape what job the agent starts, and a model's handling of them (including when it asks a clarifying question versus acting over-eagerly) is part of what the channel's vibe checks evaluate. - Slash commands are typed, slash-prefixed inputs that trigger a built-in tool action instead of a free-form prompt. - They matter in the channel's model comparisons because agent workflows — where commands direct real jobs — are a standard part of every vibe check. - Judging a model includes how it behaves at this human/tool boundary, not just its raw answers.