MediaThrive MCP connects an AI assistant to your MediaThrive team. Your assistant can inspect existing media, prepare a script and video draft, generate narration, and render the prepared video using your team’s resources.
Pilot availability: setup requires MediaThrive to enable the service and register your client. The production endpoint is
https://api.mediathrive.com/mcp. A published guide does not mean your client connection or paid generation is enabled. MediaThrive is not currently listed in the Claude or ChatGPT directories.
Choose your assistant
- Claude web, Claude Desktop and Claude Code
- ChatGPT and Codex
- Credits, permissions and team security
- Troubleshooting
Before connecting
You need a MediaThrive account, membership of the team you want to use, and the relevant MCP permissions. Connect through the MediaThrive sign-in screen, review the requested access, and select your team. Each connection belongs to one person and one team. To work in a different team, create another connection.
Your administrator must set a nonzero monthly and per-job MCP budget before paid generation can run. Normal MediaThrive billing eligibility also applies. Read-only work and saving supplied draft text do not start audio generation or rendering.
Try a first task
Start with a request that only reads your workspace:
Use MediaThrive to list my team’s recent news, available voices and video formats. Show me what is available before making anything.
Then prepare a draft:
Use this news item to prepare a short video. Write the narration here, save that exact text as a draft, and show me the scenes and an estimated cost.
When ready to produce media:
Generate narration for the prepared scenes using the selected voice, attach the completed audio, then render the current draft. Stay within the team’s MCP budget and return the finished media.
These are examples of requests to your assistant; they are not additional server tools. Paid work can proceed automatically within the permissions and limits you have granted. Your assistant may also apply its own approval settings.
Available tools
| Tool | What it does |
|---|---|
list_media | Lists your team’s videos, audio, scripts, saved assets or news. |
list_catalogue | Lists voices available to your team or shared video formats. |
inspect_media | Reads a video, script or audio item, including the current draft revision. |
estimate_media_cost | Quotes ordinary product credits and remaining MCP budget without reserving credits. |
create_video_draft | Saves supplied scene narration using an existing team news item and selected format. |
edit_script_scene | Replaces narration exactly and removes previously attached narration audio. |
edit_video_text | Changes text in an existing scene text layer. |
select_scene_asset | Adds B-roll or replaces a scene image/video with a ready team asset. |
attach_scene_audio | Attaches ready team audio that matches the narration and fits the scene. |
generate_audio | Generates the exact supplied narration, subject to a maximum delivered duration. |
render_video | Renders the current prepared draft using existing media. |
get_operation | Checks a paid operation and retrieves its completed result. |
The assistant discovers each tool’s current input schema when it connects. Lists are paginated. Draft changes require the current revision so an assistant cannot silently overwrite a newer edit.
What the pilot supports
The assistant supplies the script. Saving or editing it does not invoke a second writing model inside MediaThrive. Narration generation and video rendering are explicit paid stages. Rendering does not implicitly generate speech, avatars or missing media.
Start from news and assets already in your team. Arbitrary URL imports, asset uploads, publishing to external channels, source clip trimming, clip mute and rotation are not exposed by these tools. B-roll placement supports a start time and duration within the scene.
Audio input is limited to 5,000 UTF-16 code units (some characters count as two). Audio that exceeds the requested delivered-duration cap is trimmed, so choose a suitable cap and scene length to avoid cutting off narration.