Prompts and controls (Text to motion skill)

Part of the text-to-motion skill. The skill links here when your agent needs the detail.

Prompts

Describe one action as a verb phrase: “waves hello with the right arm”, “nods happily”, “bows politely”, “jumps up and down excitedly”, “looks around curiously”. The service normalizes to its training style (“An object waves …”).
Name the side and body part for asymmetric actions (“with the right arm”, “raises the left leaf”).
Keep it to one action per clip; 2 seconds fits one gesture, not a sequence.
Style words that map to motion work (“happily”, “slowly”, “tiredly”); story or appearance words do not.

Field	Default	Use
`prompts`	-	1-8 actions (each 1-300 chars); one clip each
`takes`	1	1-4; 2-3 when choosing a hero clip. Every take is billed
`cfgScale`	3	1.5-8; 5 for bigger, clearer gestures on stylized rigs; above 6 gets unstable
`lockJoints`	none	Base bones of rooted characters (plants, turrets, statues), e.g. `["Hips","Spine"]`
`idempotencyKey`	none	Stable key so a retried request is not billed twice
`dryRun`	false	`true` checks the rig and returns the `estimate` (credits) without making anything

Clip count = prompts x takes; each clip is billed. The call answers with a job at once; jobs.wait follows it to the finished clips.

Good: the named body part does the action, the base stays put, no mesh tearing.
Tipping over or collapsing on rooted rigs: add or extend lockJoints.
Barely moving: rephrase with a clearer verb, raise cfgScale to 5, or rename bones to human names (see rigging-for-text-motion.md).
Mesh warping at a joint: a weighting problem in the rig, not the motion; fix the rig.

Source: library/skills/text-to-motion/prompts-and-controls.md, rendered from summer-engine@4.3.0 on npm.