Input, Commands, and Context

The input supports normal prompts, file attachments, @ context, artifact mentions, slash commands, YouTube link actions, image generation prompts, and queued messages while a task is running.

Use @ to attach relevant context. Artifact mentions can pass the content of a previous output back into the next request, which is useful for revising a script, generating metadata from a draft, or creating a cover from a brief.

Slash commands can switch modes, trigger Skills, generate images, save memory, compact history, run build/test/lint actions, and start common workflows.

If an agent task is still running, Lumo can queue your next message and send it after the current task completes.

More context is not always better. For a script revision, attach the current draft and the brief. For a cover, attach the brief, title direction, and target platform. For subtitle cleanup, attach the SRT or transcript. Smaller, sharper context usually gives better outputs.

After Lumo creates a script, metadata pack, calendar, or subtitle Artifact, mention that Artifact in the next prompt and ask for a specific change. This keeps the revision tied to the previous output and reduces copy-paste mistakes.

A command name alone is rarely enough. Add the source, target platform, output format, and constraints after the command. For example, ask for a 60-second Douyin version from a specific draft instead of only typing /shorts-splitter.

Queued messages are useful when you already know the next instruction, such as 'after this draft, generate metadata' or 'after the calendar, create the first script'. Avoid stacking many vague instructions while a long task is still running.