Skip to main content
A skill is a specialized ability the agent loads for a task: generating video, editing by transcript, adding captions, analyzing beats. You rarely need to name one; describe what you want and the agent picks the right skill. This page lists every skill so you know what to ask for and how to call one directly.

​
Run a skill

  • Plain language: “add karaoke captions” runs the captions skill; “find three clips for shorts” runs a highlights edit.
  • Slash command: type / and pick from the Skills section of the palette, or type the name, for example /captions karaoke-style captions for the whole video. Each entry carries a tag for what it does (GENERATE, EDIT and so on) and a hint for the argument it takes; the words after the name become the instruction.
  • Type @ to reference a clip or asset inside the instruction; see Write prompts, references and commands.
Skill activity shows up as operation rows while the run is in progress (for example Check transcript-edit skill usage). Paid skills announce their cost and may ask for approval; see Review proposals and approve actions.
Slash palette open on the Skills section

​
Generation skills

Model choices and options for these skills are in AI models.

​
Editing skills

​
Analysis skills

​
Utility skills

​
Workflows

Workflows chain several skills into a complete production with checkpoints for your approval. Give them the source material and the goal in one message.
A workflow run in the chat with its task list and an approval checkpoint

​
The Director pipeline

/director runs a six-stage pipeline shown as a stepper in the chat: Concept → Structure → Draft → Refine → Polish → Export. Each stage ends with something to approve: concept options, an outline, a rough cut, then refinements and a final polish before export. Use Request changes on any plan card to steer it. Director runs work with the Storyboard for generated stills and clips; see Storyboard.

​
Save your own workflow

When a session went well, type /skillify to turn it into a reusable skill the agent can run again on the next project. Type /remember to review what the agent has memorized and clean up stale entries. Both are covered in Agent rules, memory and voice.

​
Continue