# Synthesia MCP > Synthesia connects your AI client to the full suite of enterprise video generation tools. Use it to create professional corporate avatar videos from simple text prompts, build content using predefined templates, or automatically dub existing media into new languages—all without leaving your terminal interface. ## Overview - **Category:** ai-frontier - **Price:** Free - **Tags:** ai-video, synthetic-media, avatar-generation, video-dubbing, content-creation ## Description This connector hooks your AI client up to the whole Synthesia suite of video tools. You can run these commands straight from your terminal, so you never have to leave your workflow just because you need a video asset. It treats your agent like a full-stack video studio director, letting you issue specific commands for everything from avatar creation and content localization right through to file cleanup. **Generating Your Videos: Avatars and Scripts** To make an avatar video, you first call **list_avatars**; this gives you the current roster of professional avatars available. Once you pick your star, you run **create_avatar_video**, giving it a text script, and it spits out a full video file featuring that AI persona speaking your words. You'll need to know which voices are ready for localization, so you can use **list_voices**; this returns all the available AI voice models, letting you check their language codes and properties before any dubbing. **Building from Blueprints: Templates** If you aren't starting with a script but need to follow corporate guidelines, you start by calling **list_templates**. This pulls up every video template you have access to. If you pick one, you run **get_template_details**, which gives you the specific structural rules and variable requirements for that blueprint. Once you know what it needs, you use **create_video_from_template**; this builds a finished video strictly adhering to that pre-approved corporate layout and structure. **Localization and Scale: Dubbing Media** Need content in another language? You don't re-record it manually. You call **dub_video**, which takes an already completed video and fully re-records the entire thing into a new language using AI voices you selected via **list_voices**. This makes scaling multinational campaigns fast. **Managing Assets: The Full Lifecycle** You'll need to know what videos are floating around or if your render is done. To see everything stored, you call **list_videos**, which gives a directory of all previously created or saved video IDs in the account. If you have a specific ID and want to check its status—like seeing how long it took to render or checking its metadata—you run **get_video_details**. When you're done with an asset and need to clear space, you use **delete_video**; remember this action is permanent and can't be undone. This setup lets your agent handle the whole media production pipeline. You manage avatar selection using **list_avatars**, check template requirements with **get_template_details**, build content with **create_avatar_video** or **create_video_from_template**, scale it globally via **dub_video**, and keep track of every file's status with **get_video_details** and **list_videos**. You own the entire process, from concept to final delete. ## Tools ### create_avatar_video Generates a complete video file using an AI avatar and the text script you provide. ### create_video_from_template Builds a video that adheres to the structure and layout of a specific, pre-approved template. ### delete_video Permanently removes a Synthesia video asset. This action cannot be undone. ### dub_video Takes an existing finished video and re-records it entirely into a new language using AI voices. ### get_template_details Retrieves the specific structural rules and variable requirements for a named template. ### get_video_details Fetches the current status, duration, and metadata for a given video ID. ### list_avatars Returns a list of all active AI avatars available to use in your videos. ### list_templates Queries and returns a directory listing of all video templates you can utilize. ### list_videos Retrieves a list of all previously created or stored videos in the account. ### list_voices Returns available AI voice models, listing their language codes and properties. ## Prompt Examples **Prompt:** ``` Give me a list of internal AI video templates, select one related to 'customer support', and execute a creation pass using proper test inputs. ``` **Response:** ``` Accessing Synthesia directories ('list_templates')... Found template 'Support Starter' (ID: tmpl-1234). Fetching target blueprint variables natively ('get_template_details'). Dispatching 'create_video_from_template' with payload values successfully... Video task submitted. The ID is 'vid-5678'. Let me know when you want to execute 'get_video_details'. ``` **Prompt:** ``` List the available AI avatars, focusing on professional corporate styles. ``` **Response:** ``` Querying Synthesia avatar registry ('list_avatars')... Here are some of the active top corporate avatars available for rendering: 1. 'Anna_Business' (ID: 101) - Blazer, formal. 2. 'Marcus_Exec' (ID: 105) - Suit, authoritative. 3. 'Elena_Office' (ID: 112) - Smart casual, friendly. ``` **Prompt:** ``` Check the status of my recent video task 'vid-9920'. ``` **Response:** ``` Evaluating task bounds ('get_video_details')... The render for 'vid-9920' is COMPLETE. Total duration: 45 seconds. You can download and stream the final file here: `https://synthesia.io/stream/vid-9920_xyz`. ``` ## Capabilities ### Generate Avatar Videos Runs `create_avatar_video` to generate a full video track using an AI avatar and a script you provide. ### Build from Templates Uses `create_video_from_template` to create videos that follow strict, pre-defined corporate blueprints. ### Duplicate Media into New Languages Triggers AI localization using `dub_video`, taking an existing video and re-recording it in a different language. ### Inventory Available Models Calls `list_avatars` to get a current list of all available professional avatars for your use. ### Manage Video Statuses Checks render progress and metadata using `get_video_details` for any existing video ID. ## Use Cases ### Onboarding a new global team The L&D lead needs 10 training videos for different regions. Instead of creating them one by one, they prompt their agent: 'List all available avatars, use the corporate template, and generate a welcome video in Spanish.' The agent runs `list_avatars`, calls `create_video_from_template`, and then uses `dub_video` to handle the localized output, solving 10 manual workflows in one chat session. ### Updating evergreen marketing content The marketing team needs to update a video script that was originally recorded in English. They use their agent to first run `list_voices` to confirm the French voice model, then call `dub_video` on the old asset ID, generating a new, localized version ready for immediate deployment. ### A/B testing video scripts The comms engineer wants to see if 'Avatar A' or 'Avatar B' works better. They prompt: 'List avatars and then run two videos using the same script, one with Avatar 101 and one with Avatar 105.' This lets them compare visual outputs generated via `create_avatar_video` side-by-side without manual rendering. ### Cleaning up old assets The project manager is running low on storage. They use the agent to run `list_videos`, identify all drafts older than 3 months, and then execute `delete_video` for each ID. This keeps their asset library clean without navigating complex web menus. ## Benefits - Cut down on manual video prep. Instead of manually logging into a separate platform, you use `create_avatar_video` directly in your terminal chat to generate assets based only on text input and avatar selection. The entire process stays contained within the AI interface. - Scale content for global teams instantly. Need to translate a training module? Use `list_voices` to check language support, then call `dub_video`. This handles localization much faster than traditional post-production workflows. - Enforce brand consistency using templates. Don't guess how the video should look. Use `create_video_from_template` and feed it a JSON payload, guaranteeing every output follows your defined corporate blueprint. - Manage assets without friction. You don't have to dig through menus just to check if a render finished. Call `get_video_details` with an ID, and you get the status immediately, letting your agent know when it’s ready for download. - Full visibility into models. Need to pick the right presenter? Use `list_avatars` to see all available avatars before starting production, ensuring the tone matches the message. ## How It Works The bottom line is that your AI client handles the sequence of API calls, letting you run complex video production workflows using simple natural language prompts. 1. Append the Synthesia MCP module into your Vinkius connective logic. You'll need to provide your specific account authorization key (`SYNTHESIA_KEY`) in the host configuration. 2. Prompt your agent with a multi-step command, for example: 'List all avatars, pick one that looks professional, and generate a video saying X in Y language.' 3. The agent calls the necessary tools—like `list_avatars`, then `create_avatar_video`—and reports back the resulting video ID. You can then use `get_video_details` to track completion. ## Frequently Asked Questions **How do I find out what avatars are available using list_avatars?** Run the `list_avatars` tool. It returns a structured list of all active AI presenters, including their specific IDs and descriptions (e.g., 'Blazer, formal'). This lets you pick one for your script. **Can I use the same video template multiple times? What tools do I need?** Yes. First, run `list_templates` to get the available IDs. Then, call `create_video_from_template`, passing the ID and your custom JSON payload to generate a new version. **If I update the script, do I have to start over? How does dub_video work?** No. You don't re-record everything. If you change the language or text, use `dub_video`. It takes an existing video ID and handles the localized re-recording process for you. **What should I do if my video render fails? Should I check status with get_video_details?** Yes. If a job stalls, use `get_video_details` and provide the ID. This tool checks the current status of the render—it tells you if it's 'PENDING,' 'COMPLETE,' or if there was an error. **How do I make sure my video uses a specific avatar? Which tools are involved?** You first run `list_avatars` to get the ID. Then, you pass that ID and your script into the `create_avatar_video` tool call. **If I need to clean up old content or remove a video, how do I use the `delete_video` tool?** The process permanently removes the asset immediately. Because this action is irreversible, you must first confirm the video ID and understand that calling `delete_video` deletes it from your account entirely. **How do I get a full inventory list of all my existing media using `list_videos`?** The tool returns a comprehensive index of every video ID associated with your account. This is essential for tracking, auditing, or referencing assets that aren't currently in active development. **Before I generate content, how can I validate the structure of my chosen layout using `get_template_details`?** It returns a detailed manifest of the template, including all necessary blueprint variables and required JSON payloads. This allows you to verify that your input data matches the structural requirements before attempting video generation. **How long does video rendering take?** The `create_avatar_video` command submits the job instantly. Rendering typically takes 2–10 minutes depending on length. Use `get_video_details` to check progress and retrieve the final URL. **Can I use my custom avatars?** Yes. The `list_avatars` tool returns both stock and custom avatars tied to your account. You can use any avatar ID when creating videos. **Which Synthesia plan is required?** API access requires a Creator or Enterprise plan. Free and Starter plans do not include API keys. Check the Integrations tab in your Synthesia dashboard to verify eligibility.