> ## Documentation Index > Fetch the complete documentation index at: https://docs.contenthero.ai/llms.txt > Use this file to discover all available pages before exploring further. # Avatars, voices and Kling elements > The identity layer: avatars with their looks, the voices they speak with, and the Kling elements that keep a character, location or prop consistent across Kling generations. The identity layer: avatars with their looks, the voices they speak with, and the Kling elements that keep a character, location or prop consistent across Kling generations. 12 tools. Every tool below is exposed by the hosted MCP server at `https://mcp.contenthero.ai` and by the `@contenthero/mcp` npm package. A write tool changes your account; a read tool does not. ### `list_avatars` read List the account's avatars. Each avatar has an imageUrl (its base look) and a defaultVoiceId, which feed generate\_lip\_sync. Call get\_avatar for full detail and the avatar's looks. The previous page's cursor. How many to return. ### `get_avatar` read Get one avatar by id: its base image (use as generate\_lip\_sync imageUrl), default voice, traits, and its looks (outfit variations). The avatar id from list\_avatars. ### `create_avatar` write Create a reusable character and start generating its first look. SPENDS CREDITS (pass getCost to preview the price without creating anything). Returns as soon as the record exists: the avatar is NOT usable yet, it sits at status 'processing' with no image until its first look finishes, so poll get\_avatar until status is 'completed'. Supply referenceImageUrls to make the avatar a likeness of a real person from their photos; omit them to invent a character from the description and traits. Apparent age, e.g. '20s', '35', 'middle-aged'. Required: the prompt writer describes the character from these traits. Gender presentation. Required, same reason as age. Avatar name, at least 3 characters. A voiceId from list\_voices, used as this avatar's default voice. Free-text description of the character. The strongest single input when no reference photos are given. Optional ethnicity, for a more specific likeness. Return the credit cost estimate instead of creating (nothing runs, nothing is charged). Content niches this character is for, e.g. \["fitness","nutrition"]. Photos of a REAL PERSON to anchor identity to: each a URL or a previous output id. Only use photos of someone who has agreed to being cloned. Visual style hint for the portrait, e.g. "editorial", "cinematic". Not stored on the avatar. ### `update_avatar` write Update an avatar and/or change its looks. Fields: name, defaultLookId (also becomes the avatar's profile photo), defaultVoiceId. Looks are changed through ops, the same shape update\_timeline and update\_canvas use: add\_look files images the account ALREADY OWNS onto the avatar, remove\_look trashes one (recoverable for 30 days). Ops run before the fields, so one call can add a look and make it the default. To GENERATE a new look instead of filing an existing image, call generate\_image with avatarId. The avatar id from list\_avatars. Place it immediately after this item of the same list. Place it immediately before this item of the same list. A look id from get\_avatar. Becomes the avatar's default look AND its profile photo. A voiceId from list\_voices, or null to clear it. New name, at least 3 characters. Look changes, applied in order before the field updates. NOT a transaction: a failure part-way leaves earlier ops applied. Place it at an end of the list instead of beside an item. Name neighbors or a position, not both. ### `delete_avatar` write Delete an avatar. Soft: the avatar stops appearing, but ITS LOOKS SURVIVE as library images and can be filed onto another avatar with update\_avatar's add\_look. Use this to retire a duplicate or an abandoned character, after moving any looks worth keeping. The avatar id from list\_avatars. ### `list_voices` read List the account's saved voices (favorites first). Set favorited=true to show only favorites. Each has a voiceId for generate\_lip\_sync / generate\_audio (TTS). Call get\_voice for full detail. The previous page's cursor. Only favorited voices. How many to return. ### `get_voice` read Get one voice by its voiceId: provider, traits (accent/language/gender/age), description, and a preview URL. The voice id from list\_voices. ### `list_kling_elements` read List the account's saved Kling elements. A Kling element is Kling 3.0's saved reference for one character, location or prop: 2-4 images (or one video) of it under a name, so the subject stays consistent across Kling video generations. Use one in generate\_video's klingElements by klingElementId, and as @name in the prompt. The previous page's cursor. How many to return. ### `get_kling_element` read Get one saved Kling element by id: its name, category, description, and images. The Kling element id. ### `create_kling_element` write Create a Kling element: a saved reference for one character, location or prop that keeps it consistent across Kling 3.0 video generations. Give it 2-4 images (or one video) of that one subject; images may be URLs or output-id tokens, so you can generate the angle shots first and assemble the Kling element from them. Use it later in generate\_video's klingElements \[\{ klingElementId }] and as @name in the prompt. What the Kling element represents (required). Referenced in the prompt as @name. Kind of entity (default 'auto'). 2-4 image URLs or output-id tokens. A single video URL or output-id token (alternative to images). ### `update_kling_element` write Update a saved Kling element's name, description, or category. The Kling element id. What kind of Kling element this is. 'auto' lets the server classify it from the image. What this Kling element is, in words. This is what makes it findable later. Rename the Kling element. ### `delete_kling_element` write Permanently delete a saved Kling element. Videos already generated with it are unaffected. The Kling element id. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.